How Repetition Can Become Overuse

Abstract

Indoor rowing is usually sold as a low-impact sport, and on a joint-by-joint basis that is broadly true. But the stroke is also a repetitive movement performed tens of thousands of times in a single long piece, and it is performed with a fixed technique whether or not the athlete is ready for it. That combination — high repetition, fixed pattern, no built-in variation — is the textbook setting for load-related problems, and the sports-medicine literature is unambiguous that load-related problems are among the most common injuries in endurance sport ([1] British Journal of Sports Medicine, Level 4).

This article reviews the evidence on how repetition becomes overuse in indoor rowing. The short answer is uncomfortable for anyone who has been told to "just row more carefully": the relationship between training load and injury is real but far weaker and more fragile than the popular 1.5 acute:chronic ratio implies. The systematic reviews that examined whether load spikes predict injury found the association to be inconsistent and heavily confounded ([8] Damsted, Level 1b; [9] Impellizzeri, Level 4), and a cluster randomised trial of load management in 482 elite youth footballers found no protective effect at all ([10] Dalen-Lorentsen, Level 1b).

What the evidence does support is narrower and more practical. Rowers' low back pain has a rower-specific mechanism, documented in a rowing-specific systematic review ([36] Nugent, Level 1b) and a 2021 international consensus statement ([40] Wilson, Level 4). Strength training is the one intervention with consistent dose-dependent protective evidence ([75] Lauersen, Level 1a). Sleep and recovery are load variables, not lifestyle variables ([68] Gong, Level 1a). And the interventions with the weakest evidence base are exactly the ones most often prescribed: screening tests ([54] Kupperman, Level 1b) and generic core-stability work ([74] Wirth, Level 4).

The bottom line: repetition does not cause overuse. Repetition under a load the tissue can adapt to is the mechanism by which you get better. The problem is not the number of strokes but whether the stroke count, the intensity, and the tissue's capacity are changing on the same timescale.

Introduction: repetition is the wrong suspect

The instinct is intuitive. A rower who rows 40 km a week for 20 weeks, feels fine, then develops pain in the lower back that does not settle, will conclude that they rowed too much. Sometimes they are right. But the same 40 km is fine for a different rower, and for the same rower it is not fine in February and is fine in May.

That variation is the whole story, and the research literature on training load and injury has spent fifteen years failing to find a clean rule that explains it. If a single number — weekly volume, a ratio, a session RPE — predicted who breaks down, the search would have ended. It has not ([1] British Journal of Sports Medicine, Level 4).

There is a second reason repetition is the wrong suspect. Repetition is what you do in the part of training that works. The adaptations that make a rower faster are overwhelmingly adaptations to repeated aerobic work ([1] British Journal of Sports Medicine, Level 4). The injuries that show up in rowers are overwhelmingly concentrated in tissues that adapt slowly relative to the training week: the lumbar spine, the tendons, the bone ([33] Kox, Level 1b; [36] Nugent, Level 1b). The mismatch is not "too much repetition". It is "the same repetition applied to a tissue that cannot remodel on that timescale".

This article is written for the rower with a connected erg and a training log, not for a clinician. It grades each claim with an Oxford CEBM level, names the places where the field actively disagrees with itself, and separates the interventions with real evidence from the ones that are merely conventional.

Methods

This review draws on 100 sources, selected with a preference for systematic reviews, meta-analyses, consensus statements and prospective cohorts over narrative reviews and single case series. Literature was retrieved from Europe PMC by PubMed identifier, with metadata verified against the live record rather than transcribed by hand. Studies in clinical populations without an exercise or sport context were excluded, as were topics outside the training-load question (paediatric, oncology, cardiac, neurological and obstetric populations). Where the evidence is a single trial or a narrative synthesis, the article says so and labels it accordingly.

The unit of analysis is the training load decision: how much, how hard, how quickly the dose changes, and how much recovery is available. That framing is deliberate. It maps onto what a rower and a coach can actually control, and it avoids the trap of treating injury as a diagnosis that appears from nowhere, rather than as the end point of a sequence of load decisions.

What "overuse" actually is

Overuse injury is not a tissue that breaks. It is a tissue that never got the chance to adapt to a demand placed on it. The clearest way to see the definition is in the surveillance studies that count injuries across a season rather than waiting for a presentation ([4] Jeffries, Level 2b; [5] Viljoen, Level 1b). In those datasets, overuse injuries dominate the injury burden, they accumulate slowly, and they are the injuries most likely to be present on the day a rower finally mentions them.

That has a direct consequence for how the evidence should be read. Studies that capture injury at first medical presentation are, by construction, studying the subset of load problems severe enough or persistent enough to seek care. That subset is not representative of the distribution the rower actually lives with ([5] Viljoen, Level 1b).

The term is also used loosely in the popular literature in a way the research literature does not support. "Overuse" is sometimes presented as meaning "too much of the same movement", when in the research usage it means "demand repeatedly applied without adequate recovery or adaptation for the tissue concerned". A rower who rows 200 km a week but changes intensity, respects rest days and has never had pain is not in an overuse situation. A rower who rows 60 km a week at one pace, never rests a tissue properly and has had the same niggle for eight months is ([14] Barry, Level 1b).

The return-to-play literature makes the same point from the other end. A 2025 consensus on return-to-play criteria after adductor-related injury treats the return decision as a criteria-based, staged process rather than a time-based one ([6] Estévez Rodríguez, Level 4). If the decision were simply "wait until it stops hurting", the consensus would not be necessary.

One further definitional point matters for the dose question that follows. Comparative work on moderate-load and high-load exercise ([7] Moreno, Level 1b) reinforces that the load distribution — how hard, how often, how far apart — changes outcomes independently of total volume. Two rows totalling the same distance with very different intensity profiles are not equivalent stimuli.

The load–injury relationship, and the ratio that broke it

No part of this field has been as influential, or as contested, as the acute:chronic workload ratio. The proposition was simple and appealing: divide this week's load by the trailing average, and a ratio above roughly 1.5 marks dangerous load. It was widely adopted in team sports and then spread well beyond the evidence base that produced it.

The research record since has been unusually blunt about this. A 2018 systematic review asked whether there is evidence of an association between changes in training load and running-related injuries, and concluded that the evidence was weak and the studies methodologically poor ([8] Damsted, Level 1b). A 2020 conceptual analysis set out why the ratio is constructed in a way that makes it statistically incoherent: the "chronic" denominator is not independent of the "acute" numerator, so the ratio is mathematically coupled to any steady increase in load, regardless of whether that increase is physiologically meaningful ([9] Impellizzeri, Level 4).

The strongest test available is the randomised one. A cluster randomised trial across 482 elite youth footballers assigned to load management or usual care found no protective effect of load management on health problems ([10] Dalen-Lorentsen, Level 1b). That is the design that matters, and it came back negative. A 2021 systematic review of the ratio's relationship with injury reported the association as present but heterogeneous across sports and settings ([3] Maupin, Level 1b), and a separate analysis found the association substantially attenuated once the statistical artefacts of the ratio's construction were accounted for ([11] Nakaoka, Level 2b).

The wider relationships between training load, sleep, and performance are similarly conditional ([12] Clemente, Level 1b). Load is clearly associated with outcomes — that direction of effect is not seriously disputed. What is disputed is whether any threshold derived from a weekly ratio is a useful decision rule for an individual athlete.

For a rower the practical consequence is direct. If a plan tells you to hold weekly volume steady for a fortnight and then add 10%, that is not a magic injury threshold; it is a conservative default, and the 1.5 number attached to it should not be treated as a physiological constant. Treat the ratio as a heuristic for plan structure and take the actual decisions from the sections that follow — tissue-specific timelines, recovery, and the interventions that do have protective evidence.

Measuring load: what the PM5 can and cannot tell you

If the ratio does not work as a decision rule, something has to replace it. The candidates are the load-monitoring methods: session RPE, distance and time, heart rate, power, stroke rate, and the distributions of all four.

The methodological literature is clear that internal and external load are related but not interchangeable, and that they can diverge substantially in exactly the situations where a load decision matters ([15] Rico-González, Level 1b). A rower can produce high external load at low internal cost on a good day, and low external load at high internal cost on a bad one. The ergometer records the first; only the rower reports the second.

Session RPE remains the most practically useful single number for a home rower, because it requires nothing but a number at the end of the session and it captures the internal cost that the machine cannot see. Its limitation is equally well documented: it is a global, unidimensional summary of a whole session, so it cannot tell you which part of the session was expensive ([14] Barry, Level 1b).

A systematic review of the relationship between training load and pain, injury and illness in competitive swimming — a sport with almost exactly rowing's structure, in that it is year-round, repetitive, technique-fixed and volume-heavy — found the load–injury evidence consistent with, but not better than, the general endurance-sport picture ([14] Barry, Level 1b). The same question has been asked in basketball with the same broadly inconclusive answer ([17] Chan, Level 1b).

Two practical implications follow. First, log more than one thing. Distance, average split, and session RPE together are strictly more informative than any one alone, and a connected erg makes all three trivial to capture. Second, be suspicious of any load metric that has not been validated for the population it is being applied to — this is a live problem in the sport-analytics literature, where algorithms are frequently applied to populations they were not trained on ([84] Claudino, Level 1b).

One constructive finding is worth knowing. A 2017 study showed that the protective association attributed to high chronic load weakens substantially once aerobic fitness is included in the model ([13] Malone, Level 2b). The rower who is fitter tolerates more load — which is the argument for building the aerobic base first, and the reason a beginner's plan and an elite's plan cannot be the same plan with different numbers.

Among the monitoring tools available to a connected rower, heart rate variability is the one most often over-claimed. The systematic-review evidence for HRV as an overtraining marker is considerably weaker than the volume of advice built on it: correlations between HRV indices and overtraining symptoms are inconsistent across studies and heavily dependent on how overtraining was defined ([16] Lipka, Level 1b). The defensible use is the same as the trend-line argument below — a multi-week rolling trend in an individual, not a single-day threshold.

Volume and dose

If the ratio is not the answer, the next question is the obvious one: is there a volume number that is safe?

The dose-response literature says that the relationship is real but that the shape depends on what you are measuring and in whom. In resistance training, volume has measurable, dose-dependent effects on strength and function, with clear dose-response relationships in older adults ([22] Ran, Level 1a). Even in the specific case of the Nordic hamstring exercise — the best-evidenced single injury-prevention exercise in sport — the volume response is not flat: higher weekly volumes produce larger protective effects than lower ones ([20] Amundsen, Level 1b).

Minimum effective volumes are well described for some modalities. Plyometric training has an identifiable minimum dose below which the adaptation does not occur ([21] Torres-Banduc, Level 1b), which is a useful corrective to the common assumption that "a bit of anything" trains everything. Resistance training volume affects functional outcomes such as gait speed measurably ([18] Nunes, Level 2b).

For the injury question specifically, the evidence is more sobering. A study of training load and injury in middle-school-aged athletes found that the load–injury relationship differed substantially by sport and that the associations were not uniform ([19] Albrecht, Level 2b). The dose that is protective in one modality is not the dose that is merely tolerable in another.

The synthesis that best fits the whole literature: there is no universal safe volume, and there is a defensible principle underneath it — the tissue with the slowest adaptation sets the pace of progression. A rower can add aerobic volume quickly, moderate tendon load moderately, and bone load slowly. A plan that adds all three at the same rate is not a well-designed plan; it is a plan that will eventually find the slowest tissue.

Tendons: the slowest tissue you own

Tendon is the clearest case of a tissue whose adaptation timeline does not match the training week. This is why tendon problems in endurance athletes are so often reported as sudden, in a rower who reports that "nothing changed" ([25] Kwan, Level 1b).

The counter-intuitive evidence here is good news. A randomised controlled trial of continued sports activity using a pain-monitoring model during Achilles tendinopathy rehabilitation found that continuing to train, guided by pain, did not derail rehabilitation ([23] Silbernagel, Level 1b). Total rest is not the answer, and the load does not have to be zeroed to be safe. The load has to be monitored and graded.

The dosing question for tendons has been examined systematically for the patellar tendon, with eccentric protocols compared against progressive tendon loading and against other approaches, with the general conclusion that heavy, slow, progressive loading is the basis of a sound programme ([24] Saithna, Level 1b). The Achilles equivalent has similarly been parameterised in terms of the variables that matter — load, volume, speed, and pain response ([26] Demangeot, Level 1b).

What the evidence does not support is a single magic protocol. Recent work on refractory patellar tendinopathy makes the heterogeneity explicit: even in athletes who have already failed simpler approaches, the label "refractory" covers a heterogeneous population requiring individualised loading ([27] Shimizu, Level 1b). A rower with anterior knee pain at 2:10/500 m and a rower with the same pain at 1:50/500 m are not the same clinical problem, even though both say "knee hurts".

Bone: slower still

Bone is slower than tendon and less forgiving, because bone adapts to load direction and magnitude rather than to effort, and because the failure mode — a stress reaction progressing to a stress fracture — is cumulative and often diagnosed late.

The clearest risk factor in rowers is mechanical, and it is rowing-specific. Rib stress fractures in rowers have a defined epidemiology, a defined mechanism and a defined management pathway ([42] McDonnell, Level 2b), and a British Rowing team guideline exists for their diagnosis and management ([93] Evans, Level 4). The rib is the archetype of a tissue that fails under repeated cyclical compression of a specific kind — the same category as the metatarsal in runners, where the risk factors have been characterised systematically ([28] Sun, Level 1b).

The systemic factors matter as much as the mechanical ones. Low energy availability and relative energy deficiency in sport are linked to bone outcomes, and the markers used to detect it are not reliably elevated in overreached athletes ([29] Gallant, Level 1a). In other words, the sign that would tell you energy availability is too low does not reliably appear — which is one reason the underlying state is easy to miss.

Vitamin D has been examined for both bone stress injury and general musculoskeletal injury risk. The evidence is consistently modest rather than absent: a systematic review of vitamin D and musculoskeletal injuries ([30] Maai, Level 1a; [94] Wyatt, Level 1a), and a meta-analysis of vitamin D and stress fracture risk in athletes and military personnel ([31] Moraes, Level 1a). The honest summary is that correcting a genuine deficiency is worthwhile and cheap, and that supplementing above sufficiency has not been shown to prevent injuries in athletes who are already replete.

The rowing-specific picture

The general endurance-sport evidence is necessary but not sufficient. Rowing has its own load profile: two daily sessions in a national squad, a year-round season, a fixed technique, and a machine that lets you produce more strokes per week than any water-based sport.

The physiological cost of that is measurable. Hormonal responses to maximal rowing before and after a heavy training block have been characterised in elite rowers ([32] Mäestu, Level 2b), which is a useful marker of how much a heavy block actually costs. The epidemiology of overuse injury in the sport has been surveyed systematically, including wrist and upper-limb distributions ([33] Kox, Level 1b).

The most useful rowing-specific work is on the association between training load and injury, and it deserves reading carefully because it is a direct test of the load-management proposition. Johnston and colleagues examined the associations between training load and baseline characteristics and injury in rowers, in two complementary analyses ([34] Johnston, Level 2b; [35] Johnston, Level 2b). The pattern that emerges is that the associations are real but modest, and that they do not reduce to a single actionable threshold — the same conclusion the general literature reached, arrived at from a sport-specific dataset.

That is the single most important rowing-specific finding in this article: rowing's own data does not support a load threshold. It supports a tissue-timeline principle.

Low back pain in rowers

If one problem defines rowers' overuse burden, it is low back pain, and it is the best-evidenced topic in this article.

The prevalence across athletes generally is high ([37] Farahbakhsh, Level 1a). The rower-specific mechanism has been examined in a rowing-specific systematic review that links rowing biomechanics to rowing-related low back pain ([36] Nugent, Level 1b). Risk factors specific to rowers have been characterised in a dedicated analysis ([41] Ze, Level 1a).

The behavioural evidence is unusually revealing. A grounded-theory study of how elite rowers actually experience and respond to low back pain found that athletes routinely downplay it — the title of the work captures the dynamic better than a summary would ([38] Wilson, Level 4). This is the most practically important finding in the whole article, because it identifies the failure mode of every load-monitoring system: the data is only as honest as the athlete's report, and elite rowers in particular are trained to report pain as weakness.

The response to a rower-specific systematic review of low back pain prevalence and biomechanics is: the spine is the tissue that suffers when the rower flexes, extends and rotates under a load the technique cannot absorb. The mechanical exposure is intrinsic to the sport. What is modifiable is the load applied to that spine, and the recovery given to it.

Management of acute low back pain in elite and subelite rowers has been addressed directly ([39] Wilkie, Level 2b), and the consensus framework for preventing and managing low back pain in elite and subelite adult rowers is the reference document for how a team should respond ([40] Wilson, Level 4). Both are consistent on the essential point: this is a managed condition with a staged response, not something to be trained through indefinitely.

Shoulders, elbows and ribs

The upper limb carries the handle through a high-repetition arc, and in rowing the load path is unusual: the arms are relatively late in the stroke, but they are the fixed end of a lever that the body must decelerate at the finish.

The best cross-sport evidence comes from swimming, which shares the year-round, volume-heavy, technique-fixed structure. A systematic review of swim-training volume and shoulder pain across the competitive lifespan found a clear relationship between training volume and shoulder pain ([43] Feijen, Level 1b). That is the closest available analogue to the rowing question and it points the same way: shoulder problems in repetitive sports are a volume-and-recovery problem, not a technique problem, though technique changes how much load reaches the tissue.

Progressive resistance exercise, with manual therapy added, has been shown effective in upper-limb overuse presentations ([44] Sharma, Level 1b). The message is the same as for the tendon: load it, progressively, with a monitorable symptom. The coach's role — and the evidence here is about the coach's role specifically — is to set that progression rather than to tell an athlete to stop ([45] King, Level 2b).

Large-scale championship surveillance puts the whole in context. Injury and illness surveillance at a major multi-sport championship quantified the musculoskeletal burden across elite competitors ([46] Edouard, Level 2b). Surveillance studies of this kind consistently show that overuse problems dominate by count while acute trauma dominates by severity and by the attention they receive. The rower's real risk is the one that accumulates quietly.

Hamstrings and the sprint row

Hamstring problems in rowers appear when the sport's dominant demand — a long, low-intensity, high-repetition cycle — is joined by a sudden high-speed demand. A 2K piece, a sprint test, or a sudden increase in rate exposure is where that surfaces.

The prevention evidence is the strongest in sport medicine, and it is strength-based. A targeted strength-training programme has a recognised potential to reduce hamstring injury ([47] Brown, Level 1b). A 2023 review of evidence-based hamstring injury prevention and risk factors synthesised the field and identified the eccentric, high-velocity component as the key element ([48] Rudisill, Level 1a). A meta-analysis comparing different eccentric exercise programmes found that the programmes differ meaningfully in their protective effect, which means "do some eccentric work" is not a sufficient prescription ([49] Hu, Level 1a).

Once a proximal hamstring tendinopathy is present, the treatment literature emphasises individualised loading rather than a fixed protocol ([50] Rich, Level 2b). The through-line across all four is that this is a load problem with a load answer, and the answer is calibrated to the individual rather than copied from a protocol.

Knees

Knee problems in indoor rowing concentrate at the front of the knee, where the repeated high-force knee extension against a fixed damper setting delivers a load that scales with rate and split.

The most useful finding for a rower here is not a treatment protocol but a prevention result: adherence to strength training is associated with lower rates of sports injury ([51] Chen, Level 1a). This is the second time in this article that the same conclusion has emerged from a completely different literature, which is worth noticing — strength training has now appeared as the protective intervention in the hamstring literature, the general injury-prevention literature, and now in the knee literature.

For patellofemoral pain specifically, conservative treatment has been compared across approaches, and the load-management elements of those programmes are consistent: progressive rather than provocative loading, and patient- or athlete-directed progression ([52] Morri, Level 1a).

A systematic review of prevention strategies for lower-extremity injury reaches the same overall shape as the strength-training literature: the interventions with consistent support are progressive strength and conditioning programmes, while the ones that fail to replicate are passive and isolated-modality approaches ([96] Bullock, Level 1a). The pattern is now consistent enough across sports to state as a general finding rather than a rowing-specific one.

Screening: what predicts and what does not

Pre-season screening is one of the most widely used and least well-supported interventions in injury prevention. It is worth being precise about why.

The core problem is a statistical one. To use a screening test clinically, it needs to be accurate enough to change a decision, and it needs to be applied to a population where the outcome is common enough for a positive result to mean something. Most screening batteries in sport have been evaluated against a base rate of injury that makes even a moderately accurate test produce more false positives than true positives. The systematic-review literature on global-positioning-derived workload metrics and injury risk makes the predictive-validity problem explicit for modern load-based screening ([54] Kupperman, Level 1b).

That is not an argument against measurement; it is an argument against using a measurement as a gate. The distinction matters because a screening result that is used only to inform a training plan is harmless, while one used to exclude someone from training is both unjust and, in a recreational sport, usually wrong.

Where screening does have support is in the older-adult context, where expert consensus explicitly recommends it and frames it correctly as part of a broader assessment rather than a pass/fail test ([53] Hughes, Level 4). Movement-quality interventions tested in controlled conditions have shown effects on functional movement outcomes ([56] Maleki, Level 1a), and visual-training interventions affect reaction time and movement ([55] Luo, Level 1a) — these are genuine effects, just not screening evidence.

The most defensible screening tool available to a home rower is not a test at all. It is a trend line: the same warm-up protocol, the same session, the same question, recorded weekly. The value is in the slope, not the absolute value.

Return to the erg

Coming back after time off, or after a diagnosed injury, is a load decision like any other, and the evidence says it should be criteria-based rather than time-based.

That principle is applied across the return-to-sport literature, from upper-limb surgery ([57] Kemler, Level 1a) to field-event athletes ([58] Wallace, Level 1a) to general postoperative rehabilitation ([59] Phelps, Level 1a). The specific criteria differ by tissue; the shared structure does not. Return is staged, criteria are objective where possible, and the last step — full training load — is a separate decision from the last step of rehab.

Biomechanical assessment tools for injury risk prediction and rehabilitation are catalogued in a way that is useful for a self-coached rower, because it distinguishes between what can be measured in a gym and what genuinely requires a laboratory ([60] Alahaidib, Level 1a).

The practical version for a home erg is simple and follows directly from the rest of this article. Return by adding one stress at a time, in the order the tissue timelines dictate: first volume, then rate and intensity, never both at once. A return plan that restores distance and intensity simultaneously is a spike, and this article's opening sections are the argument against it.

Pain, mental fatigue and the volume ceiling

There is a second ceiling on training volume that is not physical: the brain's willingness to produce maximal effort. Mental fatigue reduces training volume in resistance exercise in controlled settings ([61] Queiros, Level 1b). The effect is small but real, and it is the reason a session can be nominally identical and functionally different.

Mental fatigue has been studied across sports, including a systematic review in golf that synthesised the markers and mechanisms ([63] Pan, Level 1b). It has also been framed explicitly as a health risk factor in its own right rather than only a performance one ([64] Díaz-García, Level 4). In rowing terms, the practical reading is that the last two strokes of a long piece are not a mechanical readout of the rower's capacity — they are a readout of what the rower had left, and that includes central drive.

The wider literature on managing chronic musculoskeletal pain includes digital and hybrid self-management approaches, which is relevant to a rower managing a long-running niggle alone ([62] Villar-Alises, Level 1a). The consistent message across this literature is that the behaviours which determine outcome are adherence behaviours, not treatment choices.

The practical implication is that central drive is part of your load, and a nominally identical session can be functionally different. That is an argument for a program that survives: a short list you actually do, with mental fatigue treated as a real cost to be managed rather than as a character flaw.

Sleep, recovery and detraining

Sleep is the most under-rated load variable in recreational endurance sport, and the evidence base has strengthened considerably.

A systematic review of how sleep affects recovery and performance in a structured team-sport context sets out the mechanisms: sleep loss impairs recovery from training, and the cost compounds across consecutive nights ([66] Ochoa-Lácar, Level 1b). A meta-analysis of acute sleep deprivation on sporting performance quantifies the size of the effect across sport types ([68] Gong, Level 1a), and a consensus on sleep deprivation and perceived exertion finds the perceived-effort cost is larger than the objective performance cost ([95] Kong, Level 1a). Sleep interventions have their own systematic review, with the honest conclusion that they improve outcomes but adherence to them is the limiting factor ([67] Cunha, Level 1a).

Two modern lifestyle exposures are directly relevant to a rower who trains at home in the evening. Evening smartphone exposure degrades sleep quality and next-day performance in a controlled trial ([99] Dridi, Level 1b). That is an unusually practical finding: the device next to the erg is a measurable load variable.

The female-athlete literature reaches the same conclusions on training load in a different population, with the added dimension of load management across the menstrual cycle ([65] Costa, Level 1b). For a rower training through a pregnancy, or managing a chronic condition, the standing advice is to individualise and involve a clinician rather than apply a generic protocol ([2] NHS, Level 4).

Detraining is the control condition most plans never collect. The training-and-detraining effects of dynamic warm-up programmes illustrate the shape of the problem: adaptation disappears on a timescale measured in weeks, and the rower who stops for a fortnight does not return to the same place ([69] Iranmanesh, Level 2b). Ageing interacts with all of this, and the consensus physical-activity framework for older adults is the appropriate starting point for a masters rower rather than any training-load rule written for younger athletes ([97] O'Donovan, Level 4).

Concurrent training: the interference effect

Adding strength work to an endurance programme creates a genuine physiological conflict, and this is the one part of the literature where the disagreement is methodological rather than substantive.

The interference effect is real in principle: concurrent training can attenuate adaptations to one modality relative to training that modality alone. Four recent syntheses converge on a more useful version of the same finding. Comparative efficacy of concurrent training types on lower-limb outcomes favours particular orderings and pairings ([70] Chen, Level 1a). The broader question of concurrent versus single-modality training has been examined across outcomes ([71] Khalafi, Level 1a). Whether sprint interval training specifically causes interference has been addressed directly, and the answer is conditional rather than categorical ([72] Ferraro-Farro, Level 1a). A systematic review of concurrent training's effects on maximal oxygen consumption is the most directly relevant to a rower, and it is the reason the strength-and-endurance combination is worth the coordination effort rather than avoiding ([73] Zhao, Level 1a).

The practical synthesis, consistent with the rest of this article: separate the modalities in time where you can, do the higher-priority adaptation fresh, and treat the strength session as a load with its own progression rather than as an add-on. The coordination cost is real; so is the protective evidence for strength work ([75] Lauersen, Level 1a).

What actually reduces injury

Having spent most of this article on what does not work, the section that matters most is the one that does.

One result in the load-and-injury literature stands apart: strength training is effective, dose-dependent and safe for preventing both acute and overuse injuries ([75] Lauersen, Level 1a). The meta-analysis pooled 7738 participants across six trials for a relative risk of 0.338, and a 10% increase in strength training volume reduced injury risk by more than four percentage points ([75] Lauersen, Level 1a). This is the rare intervention where the evidence is a meta-analysis of randomised trials, the effect held across robustness tests, and the mechanism is plausible.

The practical detail that most often gets lost is the dose. The effect is dose-dependent ([75] Lauersen), which is why "do some strength work" underperforms "do a specific amount of progressive strength work". Reporting standards for resistance-training programmes have been criticised for exactly the reason that makes this hard — programme variables are reported inconsistently, so dose is not comparable across studies ([76] Goff, Level 1b). A rower following a plan benefits from knowing that the weekly sets and progression are the intervention, not a detail.

Aerobic fitness has its own protective story, and an overview of systematic reviews of aerobic fitness and training outcomes places it alongside strength work rather than below it ([77] Eitivipart, Level 1a).

Core stability deserves its own correction, because it is the most over-prescribed intervention in rowing. A critical analysis of core-stability guidelines concluded that the evidence base is far weaker than the clinical popularity implies, and that the term covers interventions too heterogeneous to support a general recommendation ([74] Wirth, Level 4). That is a claim about prescriptions, not about anatomy: core weakness is associated with injury, but no specific core programme has been shown to prevent it. Strength work is worth doing for the strength-and-injury reasons above; it is not worth doing on the promise of a stronger core preventing back pain.

Core stability, prehabilitation and warm-ups

The prehabilitation literature is best understood as a field that has largely not delivered on its promise. Prehabilitation before surgery has been reviewed systematically, and the effect sizes for the outcomes that matter are small and inconsistent ([79] Kann, Level 1b). Prehabilitation is a reasonable thing to do before a known stressor; it is not a substitute for a training plan.

Warm-ups are a better-supported and more modest intervention. A meta-analysis of personalised warm-up strategies in adult athletes found benefit that is real but modest, and conditional on the warm-up actually being specific to the activity ([81] Xu, Level 1a). Soft-tissue techniques combined with exercise have been examined for their effect on recovery and soreness, with the general finding that the exercise component carries most of the benefit ([80] Ragone, Level 1a).

This matters for a rower because a warm-up is the cheapest load-management intervention available and the easiest one to drop when time is short. The evidence does not support a dramatic claim about it. It supports the modest claim that a specific warm-up is worth the ten minutes.

Core stability is worth separating into two questions, because the evidence answers them differently. As an intervention, the guideline analysis found that most prescribed specifications have never been tested for effectiveness or compared against real strength-training loads ([74] Wirth, Level 4). As a risk factor, the picture is the opposite of what the coaching folklore assumes: a systematic review of nine studies found core strength, proprioception and neuromuscular control to be associated with lower-extremity injury, with conflicting evidence only for core endurance, and concluded that impaired core stability should be considered when screening athletes ([78] De Blaiser, Level 1b).

So the honest position is narrower than either camp states. Weak core function is associated with injury, which is not the same as showing that a core programme prevents it — association in observational screening studies is weaker evidence than an intervention trial, and the guideline review found the prescriptions largely untested. If a rower does core work, the reason to expect benefit is that it is a sensible general-purpose strength practice, not that a specific prehab protocol has been shown to prevent injury.

Adherence: the weakest link

Every intervention in this article has an adherence problem, and for injury prevention the problem is severe enough to be the finding rather than a caveat.

Two studies make the point quantitatively. A meta-analysis of adherence to exercise-based injury prevention programmes found that adherence is the dominant determinant of whether a programme works ([83] Viiala, Level 1a). A large analysis of temporal trends in adherence to aerobic physical-activity guidelines shows how little of the recommended activity population actually accumulates ([82] Abernethy, Level 1b).

The rower-specific translation is direct. The most effective injury-prevention programme in the literature is a specific, dose-dependent strength routine ([75] Lauersen, Level 1a), and the strongest predictor of whether it protects you is whether you do it ([83] Viiala, Level 1a). A plan that assumes three strength sessions a week and delivers none is not a strength-training intervention; it is a plan that has silently converted its best evidence into nothing.

This is the most actionable sentence in the article. Design the plan you will actually follow, then make that plan good — not the reverse.

What a session can tell you about technique

The connection between load data and technique is where the ergometer is genuinely useful, because a rowing-specific review of biomechanics, physiology and hydrodynamics sets out the measurable variables that respond to technique changes ([86] Yusof, Level 1b).

Spinal and pelvic kinematics during prolonged rowing on an ergometer are directly measurable and directly relevant to the low back discussion above ([85] Trompeter, Level 2b). The finding that prolonged erg rowing produces measurable spinal kinematic change is the mechanistic link between "long steady row" and the lumbar load that the consensus statement is written about ([40] Wilson, Level 4).

On-water biomechanical assessment has been reviewed as a scoping review, and it makes a useful distinction for a home rower: the variables that matter can be screened at low cost, but the ones that predict injury are not the ones you can measure at home ([87] Legge, Level 1b).

The algorithmic-analysis literature is the one to treat with most caution. The use of artificial intelligence for injury risk assessment and performance prediction is an active and largely unvalidated field ([84] Claudino, Level 1b). A connected erg can produce a great deal of data, and none of that data is a diagnosis.

Fueling and energy availability

Bone is the tissue that fails when energy availability is chronically too low, and the detection problem is that the warning signs do not reliably appear. Markers of low energy availability in overreached athletes have been examined and found not to reliably discriminate ([88] Kuikman, Level 1a). Experimental manipulation of energy availability in male athletes has been carried out, with the expected effects on endocrine and bone markers ([89] Sim, Level 1b).

Nutritional education interventions aimed at athlete energy availability have been systematically reviewed, and the finding is that education alone changes knowledge more reliably than it changes behaviour ([90] DeJong Lempke, Level 1a).

Carbohydrate availability interacts with all of this. Periodised carbohydrate restriction in endurance athletes has been examined for its performance effects ([91] Gejl, Level 1a), and the ergogenic effects of acute carbohydrate availability are well characterised ([92] Ramos-Campo, Level 1a). For an indoor rower the practical version is narrow: a rower doing two sessions a day is doing a double session with a low total energy expenditure, and the carbohydrate question is usually about the second session rather than the first.

Ageing, environment and the remaining variables

Two rowers who are both training well can still diverge, for reasons that are not about the training plan.

The evidence base for older athletes is real but narrow, and it is worth knowing what it does and does not cover. A meta-analysis of 55 studies of chronically trained master athletes found that master endurance athletes held a VO2max comparable with young healthy controls, while strength and mass are preserved in an exercise-mode-specific way ([98] McKendry, Level 1a). What that evidence does not give a recreational rower is a training-load rule: the cohorts are small, self-selected and overwhelmingly observational, so the load-injury findings earlier in this article are drawn from younger populations and applied here by inference rather than by measurement. The physical-activity consensus for older adults remains the right starting framework rather than any training-load rule derived from younger cohorts ([97] O'Donovan, Level 4).

Environment is a load variable. Heat adaptation is a genuine, trainable and sex-specific phenomenon, and the meta-analysis of heat adaptation in females is worth reading for a rower training indoors without climate control ([100] Kelly, Level 1a). The rowing-specific part is simple and frequently missed: an indoor erg in an unventilated room in summer is a heat-exposure session, and the load is not lower because the stroke is short.

The final governance point is that the guidelines themselves are only as good as the evidence they summarise. This is not a criticism of any individual guideline; it is the reason the load-and-injury literature took fifteen years to dismantle a rule that everyone had already adopted by habit ([9] Impellizzeri, Level 4).

Red flags: when to stop and get help

This section is the reason the article exists in its current form. The research literature consistently supports graduated, criteria-based management rather than training through pain ([40] Wilson, Level 4; [6] Estévez Rodríguez, Level 4), and public clinical guidance is explicit about when self-management is no longer appropriate ([2] NHS, Level 4).

Escalate to a qualified clinician rather than managing alone when any of the following is true:

• The pain is not settling between sessions, or is worsening week over week. This is the single most reliable indicator in the whole article, and it is the one that elite rowers are most likely to ignore ([38] Wilson, Level 4).
• Pain wakes you at night, or disturbs sleep persistently. Night pain is a standard clinical red flag, and sleep is already a load variable in this article ([66] Ochoa-Lácar, Level 1b).
• Numbness, tingling or weakness in the arms or legs. Neurological symptoms need assessment, not a modified plan.
• Dizziness, visual disturbance, or unusual breathlessness at rest. These are systemic, not mechanical, and they sit outside anything an erg can explain.
• Localised bone pain that is worse on impact. Rib and pelvic-girdle pain in a rower warrants assessment for a bone stress injury, which has a specific pathway in the sport ([93] Evans, Level 4; [42] McDonnell, Level 2b).
• A hot, swollen joint, or pain that is waking you from sleep to move it.
• Any new or worsening neurological symptom, or bowel/bladder change. These require same-day medical attention, not a training modification.

None of this is a diagnosis. It is the list of things that mean "stop adjusting the plan, start talking to someone". The distinction matters because the entire rest of this article is about self-management, and self-management has a boundary.

A progression that survives contact with a real life

Everything above reduces to a design rule. If you take one thing from this article, take this.

Progress one tissue at a time, in the order the adaptation timelines allow.

The order is not arbitrary:

1. Aerobic volume adapts fastest. Add kilometres early and often.
2. Intensity and rate adapt next. Add a threshold or rate block after the volume base is stable — never in the same week as a volume jump.
3. Strength work adapts on a middle timeline and carries the strongest protective evidence, so it gets its own slot and its own progression ([75] Lauersen, Level 1a).
4. Tendon, bone and the lumbar spine adapt slowest. These set the ceiling on everything else, and they are the reason a plan that adds everything at once eventually finds a limit ([36] Nugent, Level 1b; [40] Wilson, Level 4).

The four rules that follow from that:

• A plan is a sequence of single-variable weeks, not a block of simultaneous changes. This is the direct consequence of the load-spike literature being weak ([8] Damsted, Level 1b; [9] Impellizzeri, Level 4).
• Track the trend, not the number. The same warm-up, the same session, the same question, weekly. The slope is the signal; an absolute value is not ([54] Kupperman, Level 1b).
• Recovery is load. Sleep, rest days and food are variables in the same equation as volume and intensity ([68] Gong, Level 1a; [88] Kuikman, Level 1a).
• Design for the plan you will actually do. Adherence is the dominant determinant of whether any of this works ([83] Viiala, Level 1a).

And the honest caveat, which is also the finding: the randomised evidence for load management as an injury-prevention intervention is negative ([10] Dalen-Lorentsen, Level 1b). What has a positive randomised evidence base is strength training ([75] Lauersen, Level 1a), sleep ([68] Gong, Level 1a), and a sufficiently specific warm-up ([81] Xu, Level 1a). If a rower could only do three things from this article, those three are the right three.

Limitations

This review has real limitations, and they should be read before the conclusions.

The load–injury literature is dominated by observational designs, and the injury ascertainment in those studies is inconsistent — ranging from first medical presentation to self-report to medical-record coding ([3] Maupin, Level 1b; [4] Jeffries, Level 2b). This heterogeneity is the single largest source of uncertainty in the field and the reason the meta-analyses disagree with each other rather than converging.

The evidence base is also unevenly distributed across populations. Trail running, team sports and swimming are well represented ([5] Viljoen, Level 1b; [46] Edouard, Level 2b); indoor rowing on a domestic ergometer is barely represented at all. Most of what follows is an inference from structurally similar sports to indoor rowing, and that inference is explicitly labelled as such throughout.

Several of the sources cited here are recent and have not been replicated. A single systematic review is not a settled question, and the vitamin D literature in particular ([30] Maai, Level 1a; [31] Moraes, Level 1a; [94] Wyatt, Level 1a) is an area where effect sizes are modest and the underlying populations are heterogeneous.

Finally, none of the cited trials enrolled recreational indoor rowers. The most applicable evidence comes from elite rowers on the water, from national squads, and from structurally analogous sports. A rower on a home erg in a spare room is the least-studied athlete in this entire literature, and the recommendations here are correspondingly more conservative.

Conclusions

Repetition is not the mechanism of overuse. The mechanism is repetition applied faster than a slow-adapting tissue can remodel, without enough recovery to complete the adaptation between exposures.

The evidence in this article supports five conclusions and refutes one common belief.

It supports: a tissue-timeline principle over a threshold; strength training as the one intervention with a strong, dose-dependent randomised evidence base ([75] Lauersen, Level 1a); sleep and recovery as load variables with quantified costs ([68] Gong, Level 1a); a specific warm-up as a modest but real intervention ([81] Xu, Level 1a); and progressive, criteria-based return after time off rather than time-based return ([40] Wilson, Level 4).

It refutes: the acute:chronic workload ratio as a physiological constant and an injury-prediction rule. The systematic reviews found the association inconsistent and confounded ([8] Damsted, Level 1b), the ratio's construction was shown to be statistically incoherent ([9] Impellizzeri, Level 4), and the randomised trial of load management found no protective effect ([10] Dalen-Lorentsen, Level 1b).

It also refutes two quieter claims that recur in rowing coaching: that a specific core-stability programme prevents injury, which the guideline analysis does not support ([74] Wirth, Level 4) even though impaired core function is associated with injury ([78] De Blaiser, Level 1b), and that pre-season screening can gate participation, which the predictive-validity literature does not support ([54] Kupperman, Level 1b).

What a rower actually does with this: progress one tissue at a time in the order the timelines allow, log distance and split and session RPE together, treat sleep and food as load, keep the strength work because it is the one thing that reliably works, watch the weekly trend line rather than a threshold, and go and see someone when the pain does not settle between sessions — because the evidence is unambiguous that this is the behaviour most likely to cause harm ([38] Wilson, Level 4; [2] NHS, Level 4).