VO2max is the number every rower wants, and it is the one number the rowing machine cannot give them. That is not a quibble about equipment. VO2max is defined by measuring the volume of oxygen you consume against the carbon dioxide you produce, at exhaustion, through a mouthpiece. The erg has a handle and a flywheel. From those two things a great deal about your performance can be predicted — but not that quantity, and holding the two ideas together is where most of the confusion in this area comes from.
Power predicts the 2K better than oxygen uptake does
The clearest demonstration comes from the best available model of 2K performance. Ingham and colleagues tested 41 elite championship finalists and built a regression of 2K speed from four variables: power at VO2max, oxygen uptake at the blood lactate threshold, power at the 4 mmol/L blood lactate concentration, and maximal power ([1] Ingham). Together those explained 98% of the variance in speed, and the model was then validated in 18 further elite rowers with agreement limits of roughly −1.5 to +6.9 seconds.
Read the list again. Three of the four predictors are powers. The fourth is an oxygen uptake. VO2max does not appear in the model at all.
A later comparison of two standard incremental test protocols went the same way. Maximum minute power from a ramp test correlated with 2K power at r = .98; power at VO2max correlated at r = .96 ([3] Ingham). And across 76 elite rowers, the single best predictor of 2K speed was power at VO2max, allometrically scaled — which explained 95.3% of the variance in speed ([2] Nevill).
The useful conclusion is not that oxygen uptake is meaningless. It is that on an erg, the quantity that predicts your race is how much force you can put into the handle, and the oxygen figure is a step removed from that. Anyone treating a 2K as an oxygen-uptake test is looking one link upstream of the thing being measured.
The relationship curves, and that changes what "improvement" costs
There is a wrinkle that matters for anyone trying to use this for planning. Nevill's model found the fitted exponent at 0.28 with a 95% confidence interval of 0.226 to 0.334, which brackets the theoretical value of 0.33 that follows from the physics of drag rising with the square of velocity ([2] Nevill).
Read that as a coach rather than a statistician and it says something specific: the faster you already are, the more power it costs you to gain the same amount of speed. The curve is not a line, and it steepens. A rower sitting at 6:00 needs a certain extra watt to reach 5:50; a rower at 5:30 needs more than that to reach 5:20. Plateauing is therefore partly arithmetic, not purely motivational.
It also explains why comparisons across weight classes went wrong at first. Nevill's initial model could not explain sex or weight-class differences in performance until anaerobic determinants were added alongside the aerobic ones, at which point a single allometric model fitted every rower regardless of sex or class ([2] Nevill). One scaling rule does not cover everybody, and a power number does not translate across bodies without one.
Where VO2max per kilogram stopped predicting
The sharpest result for a club rower comes from Cerasola and colleagues, who tested 15 national-level youth rowers aged around 16 and modelled their 2K time against anthropometry, VO2max and 60-second all-out power ([4] Cerasola).
Every examined variable correlated significantly with 2K time except two: VO2max relative to body mass, and age. The strongest relationship by a distance was 60-second mean power, at r = −0.943. Combining anthropometry, VO2max and W60 gave R² = 0.94 with a standard error of the estimate of 6.4 seconds.
Read that carefully, because it is narrower than it sounds. These were elite youth rowers of broadly similar size, so the relative VO2max measure had little variance left to explain performance with. It is not evidence that VO2max per kilogram is worthless — it is evidence that within a group of already-selected athletes, normalising by body mass can consume exactly the signal you wanted.
Predicting a number you cannot measure
There is a legitimate way to get a VO2max estimate from erg data, and it is worth knowing what it costs. Mazza and colleagues gave 20 female club and Olympic rowers an individualised 7 × 2-minute incremental test, recorded VO2max in the laboratory, and regressed it on maximal power from the erg: r = .94, explaining 89% of the variance, cross-validated in 10 further female rowers. Mean predicted VO2max was 3480 mL/min against a measured 3530 mL/min ([5] Mazza).
Those are good numbers, and the caveats are the point. The standard error of estimate was 162 mL/min, and the percentage standard error was 4.6% — a meaningful individual error even when the group agreement looks excellent. The equation outputs absolute VO2max in mL/min rather than relative to body mass. It was developed and cross-validated in female rowers.
A prediction equation is a legitimate instrument with a stated error bar, and it is not the same object as a measurement. Anyone quoting a predicted VO2max to three significant figures has stopped reading.
Training the thing you actually care about
If VO2max is what the model is built on, the obvious question is how to move it. The best current evidence is a meta-analysis of nine studies covering 176 elite athletes, 80 of them female, comparing high-intensity interval training against conventional methods ([6] Ma).
HIIT produced a significant increase in VO2max overall, with an effect of 0.58 and a confidence interval of 0.30 to 0.87. The useful detail is in the conditions: the effect appeared when the interval recovery period was at least two minutes and the recovery phase intensity was no more than 40 percent. Those two conditions are what separated the studies where VO2max rose from the ones where it did not.
That is a more specific prescription than "add some intervals", and it lines up with everything else in this article. The stimulus that moved the oxygen number was one where a rower could actually recover enough to produce genuine high-power efforts in the next interval. Note what the finding does and does not cover: it cannot tell you that three minutes of rest is better than two, only that the studies where VO2max rose all sat at two minutes or more. Shorter than that and you are outside the conditions anyone has measured.
What the machine is actually recording
Worth being concrete, because the gap between what the erg records and what people believe it records is where the confusion starts.
The monitor times a stroke and knows how much work the flywheel absorbed. That work comes from a repeating mechanical sequence — catch, drive, finish, recovery ([8] Soper) — and within that sequence the force is not constant. In race simulations, handle force peaked at 46 percent of the stroke cycle in men and 43 percent in women ([7] Ledergerber). Two rowers can produce the same average watts while spending visibly different amounts of time at high and low force, and the machine will report the same number for both.
What arrives on the screen is therefore a derived figure: a pace, a distance, and whatever the monitor's own model converts those into ([9] Concept2). And it only means something against a stated distance, which is why the federation's standard indoor events — 500 m, 1,000 m, 2,000 m, 30 minutes ([10] World Rowing) — are the frame of reference for any comparison.
None of this makes VO2max irrelevant. It makes the erg a power instrument, and a power instrument that predicts race performance remarkably well once you stop asking it to be a metabolic one.
What to do with this article
Stop calling the 2K a VO2max test. It is the most reliable field test you have of sustained power, and describing it as an oxygen-uptake test sets up expectations that the measurement cannot meet.
Report power, not splits, and scale it before you compare bodies. Watts and split are two expressions of the same erg output — one is just the inverse of the other — so the reason to prefer power is that it is what the prediction models actually use. Raw watts still do not travel between athletes of different sizes: the allometric exponent earlier in this article is there precisely because a 380-watt rower and a 340-watt rower are not covering the same ground.
Watch the curve, not the straight line. Improvement costs more power at higher speeds, so a flat 2K at a genuinely high level is partly arithmetic. Check whether your watts are still moving before concluding you have stalled.
If you want a VO2max figure, get a real one or accept the error. A laboratory test is a measurement. A prediction equation is an estimate with a 4.6% standard error, valid in the population it was built on. Both are useful; quoting one as the other is not.
If you are adding intervals to move the number, give the work at least two minutes of recovery at no more than 40 percent intensity, and expect the sessions to feel too easy to be doing anything. That is the finding, not a flaw in it.
Key points
- In elite youth rowers, VO2max per kilogram did not predict 2K time at all. Sixty-second mean power did, at r = −0.94.
- You cannot measure VO2max on an erg. The best available option predicts it, from a population-specific equation carrying real error.
- Power at VO2max predicts 2K speed better than VO2max itself, and the relationship curves rather than scaling linearly.
- The gap between power and speed widens as you get faster, so an equal improvement costs more power at higher speeds than it does at lower ones.
- HIIT raised VO2max in elite athletes, but only when recovery ran at least two minutes at no more than 40 percent intensity.
- A 2K is a power test. Treating it as an oxygen-uptake test is what produces the wrong expectations about what is limiting it.
Editorial & Coaching Standards
Guides in the MyNextRow Learn library are created by Concept2 athletes and coaches, synthesizing peer-reviewed sports physiology, biomechanics literature, and authoritative rowing guidelines. Every workout protocol and technical cue is tested for safety, repeatability, and PM5 monitor compatibility.
Sources and further reading
- Ingham SA et al. Determinants of 2,000 m rowing ergometer performance in elite rowers. Eur J Appl Physiol 2002— Forty-one elite championship finalists; the regression model whose predictors are powers and oxygen uptakes rather than VO2max alone.
- Nevill AM et al. Modelling the determinants of 2000 m rowing ergometer performance. Scand J Med Sci Sports 2011— Seventy-six elite rowers; the allometric model showing power at VO2max explains 95.3% of speed variance and that the exponent is not 1.
- Ingham SA et al. Step-wise vs ramp-wise incremental rowing tests. Int J Sports Physiol Perform 2013— Compares the measures each test protocol yields; maximum minute power correlates more strongly with 2K power than does power at VO2max.
- Cerasola D et al. Predicting 2000-m ergometer performance from anthropometric, VO2max and 60-s power. J Hum Kinet 2020— Fifteen national-level youth rowers; VO2max relative to body mass did not correlate significantly with 2K time, while 60-second mean power did.
- Mazza OB et al. A maximal rowing-ergometer protocol to predict maximal oxygen uptake. Int J Sports Physiol Perform 2023— Female rowers only; predicts VO2max from maximal power on the erg, cross-validated, with an explicitly stated standard error.
- Ma X et al. VO2max (VO2peak) in elite athletes under high-intensity interval training. Heliyon 2023— Meta-analysis of nine studies and 176 elite athletes; the recovery-duration and recovery-intensity conditions under which HIIT raised VO2max.
- Ledergerber R et al. Strength determinants across distinct phases of Olympic rowing. Eur J Sport Sci 2023— Race simulations in fourteen adolescent rowers; the measured position of peak handle force within the stroke cycle.
- Soper C, Hume PA. Towards an ideal rowing technique for performance. Sports Med 2004— Biomechanics review of the stroke sequence; the mechanical work an oxygen-uptake figure is being asked to stand in for.
- Concept2 — Rowing Technique— Manufacturer technique guidance; the stroke the erg measures work from.
- World Rowing — Indoor Rowing— Federation reference for the indoor distances a power figure is interpreted against.