REF-721 · Langley et al., 2024 · European Journal of Applied Physiology, 2024 Apr;124 · Paper · PubMed
In actual world-class marathon racing the gain is 1.0%, about 79 seconds
A quarter of the laboratory figure
Protocol
Retrospective analysis of n=99 world-class male marathon runners across 971 performances from 2012 to 2021, split by footwear: AFT (299 performances) versus traditional (672). Year and racecourse were extracted; a mixed model for repeated measures identified athlete (p=.005) and course (p<0.001) as significant contributors to variance and included them as random factors, with footwear as the fixed factor.
Key Findings
Mean race speed 5.441 m/s in AFT versus 5.386 m/s in traditional shoes (p<0.001); mean difference 0.055 m/s (95% CI 0.039-0.071).
A 1.0% improvement — approximately 79 seconds over the marathon distance.
The authors note this is "near identical" to the degree of improvement seen in the male marathon world record run in AFT.
Their framing: 1% is smaller than laboratory models predicted, and still a substantial margin at the elite level.
Practical Application
Across 971 performances by 99 world-class male marathon runners (2012-2021), race speed in advanced footwear was 1.0% faster than in traditional shoes (5.441 vs 5.386 m/s) — about 79 seconds over the marathon, controlling for athlete and course. That is smaller than laboratory models predicted, and still substantial at the elite level.
Not from this paper: the laboratory metabolic figure was 4% (REF-719) and the pooled metabolic effect about 2-3% (REF-720); this race-speed number is the one to quote when the question is race time. The comparison crosses outcomes (metabolic cost vs race speed), so the numbers are not strictly a chain.
1% is decisive at the sharp end — dismissing super shoes because "it's only 1%" misreads what 1% means in elite racing.
Inference, not tested: if an amateur gained the same 1%, it would be roughly 2 minutes on a 3:30 marathon — worth having, not worth reorganising training around. This study says nothing about non-elite runners.
Not from this paper; the base's own practice: set expectations before purchase — an athlete told to expect 4% will feel cheated; one told to expect around a minute in an elite-level marathon is accurately informed.
Limitations
Retrospective and observational — footwear was not randomised. Athletes adopting AFT early differ from those who did not, and shoe choice correlates with sponsorship, race selection and era.
2012-2021 spans a decade of possible confounders (inference, not tested): pacing technology, rule changes, drafting, nutrition practice and the general progression of the event.
World-class males only. Nothing here about women, sub-elites or recreational runners.
Course was modelled as a random factor, but the abstract reports no control for weather or competitive context.
A 1% group estimate says nothing about the response of any individual athlete.
How this entry was checked and changed
Content correction 2026-09-27
"No control for ... elevation profile" -> course was included as a random factor in the mixed model (added with the 0.055 m/s difference and its CI).
The 3:30-marathon example labelled as inference (no amateurs studied); expectation-setting advice labelled as the base's own practice.