REF-903 · Rappelt et al., 2026 · Frontiers in Physiology, Vol. 17, Art. · Paper · PubMed · Free full text

Longitudinal performance development in PRO and ELITE HYROX competitions across the first seven competitive seasons

B+moderate evidenceMeasured on trained to elite athletesWritten before October 2026, being re-checked

Public results from individual PRO and ELITE HYROX races over the first seven seasons (2018/19-2024/25; 11,842 female and 27,854 male results) were analysed for how performance evolved and which parts of the race carry it. PRO medians improved by about 8-10 min between seasons 4 and 7, with bigger gains in the middle of the field than at the slow end.

In the Top 100 per season, total time fell by about 13 min in men and 17 min in women from season 1 to 7, mostly through faster running, which stayed at about half of race time and whose ranking closely tracked the overall ranking. Strength-determined stations reshuffled rankings more and cost slower athletes disproportionately more time.

ELITE medians dropped by several minutes and the spread between athletes narrowed from over 10% to under 5%. The authors conclude that performance is most strongly associated with running, with meaningful contributions from the strength stations.

Participants

Race results (not unique athletes): females n = 11,842; males n = 27,854; individual PRO and ELITE competitions, seasons 2018/19-2024/25.

Protocol

Publicly available race results.
Season-specific percentile curves of PRO results from empirical cumulative distribution functions.
Changes in discipline performance assessed in the Top 100 results per season.
Discipline importance via rank-based reshuffling metrics and quantile regression.
ELITE: performance trends, convergence and Top-5 retention.

Key Results

PRO, seasons 4-7: median times for men and women got ~8-10 min faster, and the 25th-75th percentiles gained more than the 90th percentile did.
Top 100, seasons 1-7: men cut ~13 min (≈19%) from total race time and women ~17min (≈21%); these gains went mostly with quicker running (males: ~8 min; females: ~10 min).
Running consistently took up ~50% of total race time, and running ranks differed little from overall ranks.
Strength-determined stations showed greater rank reshuffling and stronger quantile-dependent effects — larger absolute increases in total race time among slower performers.
ELITE median: 01:06:24 to 00:57:17 h in males and 01:11:09 to 01:03:22 h in females; the coefficient of variation fell from >10% to <5% (increasing competitive density).

Practical Application

To "what matters most for a faster HYROX?": across seven seasons of PRO and ELITE results (27,854 male and 11,842 female), running was ~50% of race time, tracked the overall ranking most closely and supplied most of the Top 100's improvement (males ~8 of ~13 min; females ~10 of ~17 min); the strength stations mattered too, and cost the slower athletes the most time. Running fitness is the first priority; for a slower PRO athlete the heavy stations are the second big lever.
The station finding is quantile-dependent: at the slow end of the field a weak strength station costs disproportionately more. Inference, not tested here: an athlete who loses a lot of time on sleds or lunges has more to gain there than a fast athlete who does not.
Not from this paper; the base's own practice: this matches the only physiological HYROX study (REF-510) in a much larger, faster population; how to split the week is in SYNTH-71, and concurrent-training scheduling in SYNTH-21.

Limitations

Race results only: no physiology, training, body composition or athlete characteristics.
Results, not unique athletes; repeat racers appear more than once.
The "importance" of a discipline is a statistical association within the results; it does not show that training that discipline would lower a given athlete's time.
PRO and ELITE only; the Open division is not analysed.
Course, venue and rule changes across seasons cannot be separated from true improvement in the abstract.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

large observational dataset (n=39696 results, the sum of 27,854 male and 11,842 female), associations only