A 5K Result: What Twenty Minutes of Rowing Actually Measures

A 5K on the erg is about twenty minutes of work, and that one fact settles most of what follows. At this duration you are not near any crossover point between energy systems. You are deep inside the oxidative regime, a long way past the point where the question stops being interesting. The useful questions are about how you distributed the work and what your result is comparable to.

Where the energy split actually comes from

The number most often attached to a piece like this — roughly 85% aerobic, 15% glycolytic — is not a measurement. It descends from tabulated models built decades ago, and the largest review of energy-system contribution ever assembled says as much indirectly: of 102 studies contributing 311 individual measurements, only 7 used theoretical models, while 66 used accumulated oxygen deficit, the method that measures the oxygen shortfall rather than predicting it ([1] Gastin & Suppiah).

That review also puts the crossover somewhere specific. Anaerobic supply predominates only until roughly 75 to 80 seconds. The duration at which anaerobic and aerobic contributions are exactly equal is 78.6 seconds, with a 95% confidence interval of about ±1 second, after which the aerobic share climbs steadily for the remainder of the effort ([1] Gastin & Suppiah). Eighteen to twenty-two minutes is a little over fifteen times that crossover.

Which is also why the percentage deserves less precision than it is usually given. The same review found small but statistically significant differences between measurement methods, and small significant effects of pacing strategy — so two studies can report different aerobic fractions for the same effort without either being wrong ([1] Gastin & Suppiah).

What rowing has actually measured

The rowing-specific numbers are thinner than the general ones, and every direct rowing measurement in this literature is of a 2,000 m piece or shorter. In race simulations in eight rowers, the aerobic, phosphocreatine and glycolytic shares on the ergometer were 84%, 7% and 9%, and on the water 87%, 7% and 6% ([2] de Campos Mello). A 5K is longer than that piece, so its aerobic share is higher than 84% — but nobody has measured what it is.

The best available estimate comes from accumulated oxygen deficit rather than from a table. In eighteen junior rowers, the anaerobic contribution to a 2,000 m piece was 17.3% and to a 1,500 m piece 22.2% ([3] Astridge). Both efforts are shorter and faster than a 5K, so both overstate what a 5K would involve. Extrapolating along the duration axis is sound in direction and unreliable in magnitude.

That study produced a second result worth having on its own. The 1,500 m piece produced higher mean power than the 2,000 m — 341 against 324 watts — and at the same time lower heart rate, lower perceived effort and lower lactate ([3] Astridge). Shorter, harder, and physiologically calmer simultaneously. If you have been told that a faster piece must be a more uncomfortable piece, the measurements do not say that.

What your split predicts, and what it does not

The strongest performance-prediction work in rowing is 2K work. Among 41 elite finalists, a model combining power at VO₂max, oxygen uptake at the blood lactate threshold, power at 4 mmol/L lactate and maximal power explained 98% of the variance in 2,000 m ergometer speed, and it held up in a further 18 elite rowers to within about seven seconds ([4] Ingham). That is an excellent model of a 2K. It is not a model of a 5K, because none of the testing behind it was performed against one.

Newer evidence complicates even the 2K case. In 38 well-trained rowers, critical power predicted rowing performance better than two separate lactate-threshold methods, at r = −0.81 against r = −0.75, and adding work above critical power predicted 2,000 m time at r² = 0.92 in men and 0.81 in women ([5] Dwyer). The margin is small, the methods assign about thirteen watts of different absolute power to the same boundary, and treating the two concepts as interchangeable is disputed in a published comment on that paper ([6] Dotan).

One recent study does include a genuine 5,000 m measure. In 17 elite female rowers, squat, deadlift and power-clean one-repetition maxima correlated with average power over 5,000 m at r = .56 to .86 ([7] Behm). Strength clearly travels with the number. But it is cross-sectional in 17 athletes, so it shows the two move together, not which one to move first.

Pacing: the data says something else

Elite pacing is usually described as even or negative. The largest analysis of its kind — split times from 1,612 championship race profiles, 948 of which met the inclusion criteria — found something different. Rowers held the first 500 m at 103.3% of their average speed on the water, then settled to 99.0%, 98.3% and 99.7% for the sectors after it ([8] Garland). On the ergometer the shape was similar but gentler: 101.5%, then 99.8%, 99.0% and 99.7%.

So the profile is a fast start followed by sectors running slightly slower than average — which is not a negative split, whatever the results sheet looks like. And the more useful finding is the second one: there were no differences in pacing profile between winners and losers, and none between men and women ([8] Garland). Everyone did broadly the same thing, and the placing came from somewhere else entirely.

Read on its own that is a caution rather than a recipe. If the medalists and the finalists paced identically, pacing is not what separates them — and a negative split is not the thing to copy out of a results table.

Polarized training: the claim that did not survive

Around 80% of training below the first threshold and 20% above the second is widely described as the empirically dominant distribution for athletes at this level. The rowing-specific evidence is weaker than that sentence implies.

The one controlled trial in national elite rowers ran eleven weeks with fourteen men, holding the low-intensity zone at about 93% in both groups so that only the balance between middle and hard differed. The polarized group did become measurably more polarized — a Polarization-Index of 3.0 against 1.9 — and it did not become measurably better. Neither group's 2,000 m power, VO₂max nor power at 4 mmol/L changed significantly, and no performance measure differed significantly between the two. The authors concluded plainly that polarized training was not superior, and suggested why: the very high and very similar low-intensity share may have left no room for the distribution to matter ([9] Treff).

A 2024 systematic review with meta-analysis reached a more nuanced version of the same conclusion. Across 17 studies and 437 subjects, polarized training improved VO₂peak — SMD 0.24, 95% CI 0.01 to 0.48, with high certainty of evidence — concentrated in shorter interventions and highly trained athletes. It did not improve time-trial performance (SMD −0.01, p = 0.92), time to exhaustion (p = 0.24), or power at the second threshold (p = 0.75) ([10] Silva Oliveira).

That distinction is the entire story. Polarized training appears to build aerobic power. On the outcomes a 5K actually produces, it has not demonstrated an advantage over other distributions.

Taking something useful from the test

A 5K is still worth running. It is long enough to be aerobic-dominant, short enough to be repeatable, and comparable to itself over time even where the physiology behind it is inferred.

Three habits make it more so. Set the machine up the way you would set it up to race: same foot-stretcher height, same damper setting. These are two different things and rowers routinely conflate them. The foot-stretcher sets leg length, so it changes with your body and with what you are wearing; the damper sets drag factor, so it changes the force the handle resists against. Move either one and you have changed the test you are comparing against. Record the stroke rate beside the split as well, because the same split at a different rate is a different effort. And repeat it three or four weeks later, because the change between tests is the signal and the absolute value is context.

What this article cannot tell you

The 5K-specific rowing literature is small enough to state plainly. There is no direct measurement of energy-system contribution at 5,000 m. The strongest performance models in the sport were built against 2,000 m, and the best of them exclude VO₂max per kilogram from their predictors entirely. The pacing evidence comes from 2K racing on water and ergometer, not from 5K. Nearly all of it comes from young, well-trained, mostly male rowers. And whether polarized training helps depends entirely on which outcome you ask about: aerobic power, apparently yes; anything you would call a result, not demonstrated.