REF-1428 · Eyestone et al., 1993 · The American Journal of Sports Medicine · Paper · PubMed

Can cycling or pool running keep my running fitness while I can't run?

In 32 healthy young men over 6 weeks, water running, cycling and running at matched dose each lost a little VO2max; none changed 2-mile time. Small, 1993, abstract.

Main caveat: Abstract only; no numbers for any outcome, no SDs, no confidence intervals.
limited evidenceMeasured on recreational to trained athletes

Eyestone and colleagues (Brigham Young University) compared water running, cycling and running for keeping VO2max and 2-mile run performance over 6 weeks.

Thirty-two trained subjects aged 18-26 were tested for VO2max and 2-mile time, stratified by the 2-mile pretest into high, medium and low performers, and randomly assigned to water running, cycling or running training; the three groups trained with similar frequency, duration and intensity.

After 6 weeks every group showed a small but statistically significant decrease in VO2max and no change in 2-mile time, with no differences by training mode or by pretest level. The authors conclude that over 6 weeks runners who cannot run because of a soft-tissue injury can keep VO2max and 2-mile performance "similar to running training" with either cycling or water running.

Our reading: the only head-to-head in the base of the two substitutes an injured runner is usually offered against running itself, in a substitution rather than an add-on design: the pool and the bike did no worse than running at the same dose.

"Maintain" is loose — VO2max fell in all three arms, the running arm included — so the shared dose was a little below what these men had been doing, or the tests drifted; the trial shows equal substitutes at that dose, not that a substitute holds a runner's full load. It is an old trial (1993).

Newer work in the base: Kwok 2022's systematic review of deep-water running includes it (REF-1437, B+); the 2026 meta-analysis of running-versus-cycling trials does not (REF-1368, A-); no trial of either substitute in injured runners exists.

Protocol

Participants: n = 32 "trained subjects" aged 18-26 (men by PubMed indexing; Kwok's extraction table: 32 men who had run 1.5 miles in under 10:45, REF-1437); VO2max and 2-mile run time tested before and after.
Allocation: stratified by 2-mile pretest into high, medium and low performance levels, then randomly assigned to water running, cycling or running. Per Kwok's extraction table, 10, 11 and 11 participants.
Training: 6 weeks with "similar frequency, duration, and intensity" in the three groups. Per Kwok's extraction table: sessions of 20-30 min, 3 a week in week 1, 4 in week 2 and 5 in weeks 3-6, 27 in all, water running in a diving pool with a flotation device. How intensity was set and checked (heart rate, %VO2max or RPE) is not in the abstract.
Outcomes: VO2max; 2-mile run time.
Funding and interests: none listed in the PubMed or Crossref record; no conflict statement in the abstract record.

Key Results

Outcome after 6 weeksWater runningCyclingRunningBetween modes
VO2maxsmall, statistically significant decreasesmall, statistically significant decreasesmall, statistically significant decreaseno difference
2-mile run timeno changeno changeno changeno difference
Pretest level: high, medium and low performers responded alike.
Effect sizes computed by Kwok 2022 from this trial ([REF-1437]): 2-mile time, water running against running and against cycling, SMD −0.41 to −0.27 (described as a similar effect); the abstract itself gives no numbers.
Not reported in the abstract: the size of the VO2max loss, any heart-rate data, running economy or longer races.

Mechanisms

Measured: a slight VO2max loss in all three arms and unchanged 2-mile time, with no difference between water running, cycling and running.
Authors' interpretation: either substitute keeps VO2max and 2-mile performance as well as running does over 6 weeks, so a runner with a soft-tissue injury can use either.
Inference: a 2-mile race leans on aerobic power, which all three modes load; it does not test running economy over longer races, durability, or the bone, tendon and impact tolerance that cycling and water running do not train (PHYS-37, PHYS-38, PHYS-22). Equal results here say nothing about how ready the legs are to run again after a substitution block.
Inference: "similar intensity" across three modes cannot have meant the same heart rate if it was set on land — heart rate runs lower in water and on the bike at the same effort (REF-1433, REF-1432); the abstract does not say which anchor was used.

Practical Application

Can cycling or pool running keep my running fitness while I can't run? In 32 healthy young men over 6 weeks, water running, cycling and running at matched dose each lost a little VO2max; none changed 2-mile time. Small, 1993, abstract.
Bike or pool: they did equally here, so choose the one that does not provoke symptoms (SYNTH-94, SYNTH-73); for bone-like pain the clinician leads. Not from this paper; the base's own practice: keep the hard sessions in the substitute (REF-587, PHYS-41).
Minutes: the trial matched frequency, duration and intensity across modes, so the tested rule is minute-for-minute at matched intensity (as in REF-922 and REF-953); the 1.5-2× bike minutes coaches use for an easy run is their practice (REF-1440, C), not what held fitness here.
Intensity in water: not from this paper; the base's own practice: set it by effort and structure with heart-rate targets lowered for the water (REF-1433); recreational runners steering by feel undershoot (REF-1438).
What it does not show: anything in injured runners — the line in SYNTH-73 that no entry tests cycling as a substitute for injured runners still holds; it now has a healthy-runner trial of cycling against running and water running. Nothing beyond 6 weeks, in women or in better-trained runners.

Limitations

Abstract only; no numbers for any outcome, no SDs, no confidence intervals.
Healthy participants: the injury conclusion is the authors' extrapolation.
Small arms (about 10-11 each); a null between arms is not proof of equivalence.
Intensity matching undescribed in the abstract (anchor, monitoring, adherence).
VO2max fell in every arm, so the common dose was below the participants' usual one or testing drifted.
Reporting, per Kwok 2022's quality table ([REF-1437]): participant characteristics and confounders not clearly described, allocation not concealed, losses to follow-up not taken into account, no power calculation.
Old trial (1993); "trained" is not defined in the abstract and Kwok's extraction suggests a modest entry standard.
Funding and interests: not stated in the record.
How this entry was checked and changed
Why this level

(from the abstract) — a randomised three-arm trial of 32 (the rubric's B+ band for a parallel RCT of 20-39) held one step lower: about 10-11 per arm, no means, SDs or confidence intervals in the abstract, healthy participants, and a null between arms that cannot show equivalence