The 1% treadmill grade is a speed-dependent correction, not a universal setting
It starts mattering at about 12 km/h, and below that 0% and 1% are indistinguishable from the road
Protocol
Nine trained male runners ran 6 min at each of six velocities — 2.92, 3.33, 3.75, 4.17, 4.58 and 5.0 m/s (10.5, 12.0, 13.5, 15.0, 16.5 and 18.0 km/h) — with 6 min recovery between runs. The set was repeated five times on a treadmill at grades of 0%, 0%, 1%, 2% and 3%, and once outdoors along a level road. Oxygen uptake was the comparison.
Key Findings
| Speed | Road VO₂ matched treadmill at |
|---|---|
| 10.5 and 12.0 km/h | 0% or 1% — indistinguishable |
| 13.5 km/h | 1% (and not 0%) |
| 15.0 and 16.5 km/h | 1% or 2% |
| 18.0 km/h | any grade tested |
The headline "use 1%" is the study's summary across the whole 10.5-18 km/h range, and it is defensible as a single default — but the underlying data are speed-dependent, and the paper does not support a claim that the correction only matters for fast runners.
The band where 1% is the specific match begins around 13.5 km/h. Below roughly 12 km/h the road sits inside the 0%-to-1% envelope, so the setting is a matter of indifference rather than a correction to skip.
At the top of the range the distinction dissolves again — at 18 km/h the road was not statistically separable from any grade tested, which is a statement about measurement variance at that intensity, not evidence that grade stopped mattering.
Practical Application
A recreational runner at 13-15 km/h is inside the band where 1% is the matched setting, not below it. Telling such an athlete that the correction is for faster runners and that 0% is fine reverses what the data show.
Below ~12 km/h, do not spend the athlete's attention on it. Either setting reproduces road cost; prescribe by the number that keeps them consistent between sessions.
Prescribe the grade once and keep it fixed across a block. The purpose is comparability — a session at 0% and the next at 2% are not the same session, and the athlete's own week-to-week comparison is worth more than matching an outdoor equivalent exactly.
Not from this paper; the base's own practice: treat treadmill calibration as the larger uncertainty. Belt-speed error on an uncalibrated treadmill can be of the same order as the metabolic difference between 0% and 1% — this study did not measure it.When indoor HR at a prescribed speed sits clearly above the athlete's outdoor HR at the equivalent pace, trust HR and RPE and adjust the speed rather than the grade — and record which surface the session was run on, because a pace from a treadmill and a pace from the road are not the same measurement.
Limitations
n=9, trained men, 1996, one treadmill. Small, single-sex, and old enough that treadmill belt and deck construction have changed; the absolute numbers are not a calibration standard for a specific modern machine.
6-minute steady-state bouts. Nothing here speaks to a long run or to interval work, where drift, cooling and boredom act on the indoor session in ways the metabolic equivalence does not describe.
Level road, no wind reported as a controlled variable in the abstract. Air resistance is the mechanism the authors offer for the correction, and this design does not isolate it, so treat 1% as a practical default rather than a settled physical account.
Statistical non-difference is not equivalence. Several of the cells above rest on failure to reject at n=9; a wider band would likely separate more of them.
How this entry was checked and changed
Content correction 2026-09-27
All speeds, grades and the per-speed matching table re-checked against the abstract (velocities converted from m/s; 2.92 m/s = 10.5 km/h ... 5.0 m/s = 18.0 km/h) — unchanged.
"Belt speed commonly deviates by a few per cent" -> relabelled as the base's own practice; the paper did not measure treadmill calibration.
Removed the unsourced claim that later field-vs-treadmill work reports economy differences this study does not explain.