SYNTH-71 · Guide
Preparing for HYROX
Build the running engine first, add heavy strength short of failure, rehearse running off the stations, and start the first kilometre slower than feels right; HYROX evidence is race-result analyses and small observational studies with no training trial, so the week, the station work and the pacing targets are practice
The question it answers
An athlete has a HYROX race (event format, not from a study: 8 × 1 km runs, each followed by a station — SkiErg, sled push, sled pull, burpee broad jumps, rowing, farmers carry, sandbag lunges, wall balls). What decides the time, how to split the week between running and stations, how to train running off a station, how to pace and fuel it, how to taper, and what gets injured.
What decides the result
Running is most of the race. In 11 recreational HYROX athletes (27% women, median 18 months' experience) doing a simulated Open Division race, the total was 86.5 (14.5) min: 51.2 min running against 32.8 min at stations (p=0.003), with 79.5% of the time at very-hard and 19.6% at hard heart-rate intensity (REF-510, B).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) (REF-903, B+; results only, associations).
The engine separated the finishers. In the same study the faster athletes had higher VO2max (p=0.01), more endurance-training volume (p=0.04) and lower body fat (p=0.03); the authors describe HYROX as endurance-emphasised with moderate-to-low demands on maximal strength, coordination and mobility compared with other functional-fitness formats (REF-510, B; cross-sectional correlations in 11 people, not a training trial).
Stations are where the peaks are. Maximal lactate (8.5 vs 7.7 mmol/L) and RPE (18 vs 16) were higher at stations than on the runs; HR, lactate and RPE peaked at the last station, wall balls; the heaviest stations, sled push (128 s) and sled pull (155 s), were the quickest (REF-510, B). In PRO/ELITE results the strength-determined stations reshuffled rankings more and cost slower athletes disproportionately more time (REF-903).
No single quality predicts functional-fitness performance. Across 21 studies of mostly recreational CrossFit athletes (transfer: CrossFit is varied, HYROX is standardised), body composition and competition experience were the most consistent influences, and total-body and back-squat strength correlated with scores in a third of studies (REF-590, B+).HIFT studies commonly reported VO2max gains of 8% to 15%, not tied to HYROX (REF-905, B-, scoping review); a narrative review calls HYROX aerobic-first (REF-906, C+).
Inference: for a Tier 1-2 athlete the default lever is the running engine; stations become the lever only when the athlete's own splits show them costing more than the runs. Percentile tables for every run and station by category, from 186,411 official results, let an athlete place each split against the field (REF-901, B+; the values are in the paper, not the entry — quote none).
Default (what to prescribe first)
For a healthy Tier 1-2 athlete 12-16 weeks from a HYROX race, 5-7 sessions a week (the counts and doses in this list are the base's own practice unless a refId is attached):
1
2
Heavy lower-body strength twice a week, stopped short of failure. Across 28 controlled studies in running, cycling, skiing and swimming (not HYROX), added strength improved middle- and long-distance performance (SMD 0.52) and maximal-force training worked better than other intensities (REF-074, A-); in 24 studies of distance runners added strength generally improved economy by 2-8%, not in every study, with VO2max typically unaffected (REF-072, A-).
The prescription is SYNTH-44.
3
One station session a week, turning into race-specific combinations (a station, then a 1 km run) in the last 6-8 weeks (practice; §Compromised running).
4
Keep the lift away from the key run. Separate strength and endurance by ≥3 h when power matters: the explosive-strength penalty of concurrent training was concentrated in same-session designs (REF-330, A, healthy untrained-to-active adults). A hard strength bout can blunt the next endurance session for "several hours to days" (REF-328, B+, narrative). Placement rules are SYNTH-31.
Decision (who → what)
| Situation | Do this | Source and how sure |
|---|---|---|
| Runner-origin athlete (good 10 km, little gym history) | Keep the running; add 2 heavy sessions a week after 2-4 weeks at moderate load to learn the lifts; one station session a week | Inference from REF-510 (B: VO2max and endurance volume went with faster times): the engine is already built; heavy strength for endurance athletes (REF-074, A-); the learn-the-lift ramp is SYNTH-44's practice |
| Gym-origin athlete (strong, runs ≤2×/week) | The engine is the limiter: grow to 3-4 runs a week before adding station volume; strength to 1-2 sessions, heavy, short of failure | VO2max and endurance volume separated finishers (REF-510, B); gradual build is a heuristic (REF-619, B, novices); in 43 trained male rowers moderate-volume lifting short of failure beat higher-volume to-failure lifting for upper-body strength, and both short-of-failure groups gained more 20-min row power (REF-640, A-, transfer; failure and volume confounded) |
| Lifts stall, strength and running must share a session | Lift first when the session is for strength (sled drive, lunges); run first when it is the key run | Lifting first gave +6.91% lower-body dynamic strength (REF-588, A, healthy adults, Tier 0-2); endurance-first when the run matters is REF-329's direction (B+, no hours) |
| Horizon < ~8 weeks, or first HYROX | Specificity, no new lever: weekly compromised-running sessions, one or two partial simulations, pacing rehearsal | SYNTH-42 horizon bands (house defaults); simulations to expose the limiter are REF-590's practical line, not a tested method |
| Pro Division or Tier 3+ | Running still first; the heavy stations are the second lever, most for the slower part of the field | REF-903 B+ (results, associations); how much sled and carry work to add is practice |
| Masters or first-time lifter | Same structure; ramp lower-body load and running on the slower system | SYNTH-44 masters and beginner rows; tendon lags muscle is the base's framing in SYNTH-21, not a finding |
Compromised running (the HYROX brick)
Running after prior exercise is slower — in triathlon. Most of 39 studies found running worse after cycling than in a control run, with no cadence or intensity trick reliably protecting it (REF-655, A-); in 9 well-trained triathletes a 3 km run was 7-9% slower after 20 min of cycling (626-638 s vs 583 s), whatever the cadence (REF-656, B). Transfer: these measure cycle-to-run, not station-to-run.
The penalty depends on the athlete. Running after cycling cost 1.6-11.6% more energy, less in better triathletes (REF-589, C+, narrative); in 7 elite triathletes neither an easy nor a hard ride changed running economy or kinematics (REF-539, B), and the abstract notes cycling does impair recruitment in some highly trained triathletes. Whether training shrinks the penalty is inference in both entries.
Station legs are local, not only cardiac. In 9 highly trained trail runners, 27-km race time correlated with a knee-extensor fatigue index (r = 0.91) and VO2max (r = −0.76), not with peak torque (REF-780, B; trail, not HYROX, correlational). Inference: sled push and lunges load the same muscle before a run.
Not from a study; the base's own practice: from ~8 weeks out, once a week, pair a station with a 1 km run at goal pace (sled push → run, lunges → run, wall balls → run are the hardest pairs), 3-5 pairs; judge it by how fast the run pace settles.
Station work
Not from a study; the base's own practice, unless a refId is named:
SkiErg and rower: treat as aerobic work — intervals near threshold, not maximal pulls. In elite rowers 2000 m erg speed correlated equally with power at VO2max, maximal power and maximal force (r = 0.95) (REF-533, B; elite rowing, transfer — force and power tie with the aerobic measure, so this does not show the erg is aerobic-only).
Burpee broad jumps: a jump dose sits on top of heavy lifting, not instead of it (SYNTH-45's default, for runners); technique under fatigue is practice.
Farmers carry: no entry — practise at race load.
How to progress / periodise
House practice: weeks 16-9 = engine and heavy strength, stations once a week for skill; weeks 8-3 = add the compromised-running session, one partial simulation every ~2 weeks (e.g. 4 runs + 4 stations); strength drops to about one heavy session every 7-10 days, load kept heavy and sets cut, inside 8 weeks of the race (SYNTH-44's in-season row).
Taper: about two weeks, cut volume hard, keep every hard session and the session count (SYNTH-34, B+; house reasoning there: shorter for a small event). Practice: last heavy sled or lunge session ≥7 days out.
Race day: pacing, fuel, fluid
Pacing: in 24 male PRO racers, running averaged 84.13% of VT2 speed and 76.33% of vVO2max, falling from 16.13 km·h-1 on run 1 (just below VT2 speed, 16.41) to 13.58 km·h-1 on run 8; the Roxzone was slowest, 10.52 km·h-1 (REF-902, B, one race; describes, does not test a strategy).The entry's inference: a little under VT2 pace is a ceiling for the early runs, and brisk transitions are cheap time. In events longer than ~2 min more even pacing may improve times (REF-010, B+, narrative); across 1.7 million recreational marathoners a fast start went with a slower finish (REF-820, B+; marathon, observational).Not from a study; the base's own practice: run the first 1 km at or slightly slower than the pace you can hold for all eight; set run splits from off-station pace, not fresh 1 km pace (the REF-589 entry makes the same point by analogy from cycle-to-run); hold a reserve for lunges and wall balls (REF-510's peak there; the reserve is its inference).
Fuel: under 90 min sits in SYNTH-02's <90 min row, 30-60 g/h carbohydrate at any level; a finish beyond 90 min moves to its 90 min-3 h row, 60-90 g/h for amateurs (the first row follows consensus, the second is base practice — SYNTH-02).Not from a study: for most athletes a carbohydrate meal 2-3 h before plus, optionally, one gel or sports drink around halfway is a practical minimum, below that row's 30-60 g/h (HYROX fuelling is untested); rehearse it in a simulation.
Fluid: no entry covers a 60-90-min indoor race. Inference: start hydrated and drink to thirst.
Sleep: in 8 experienced male HYROX athletes, on nights after a resistance session sleep latency was longer (29 vs 10 min) and wake after sleep onset shorter (31 vs 48 min) than after an endurance session (session time relative to bedtime not reported); total sleep ~6.6 h (REF-521, B, pilot).
When NOT to use it / red flags
Do not build a HYROX block around maximal strength for an Open Division Tier 1-2 athlete (REF-510: moderate-to-low maximal-strength demands).
Do not prescribe weight loss from REF-510's body-fat correlation; it is cross-sectional in 11 people.
Do not add running volume in large jumps for a gym-origin athlete (REF-619, trend only); a previous injury and a long weekly distance in male runners are the strong risk factors, the knee the commonest site (REF-564, A-, distance runners).In 89 HYROX athletes followed weekly for 12 weeks, about 30% had a health problem in a given week, injuries ran at 14.4 per 1,000 h, 73% were overuse, the knee and lower leg carried the most burden, and only half of problems caused time loss — ask about pain, not days missed (REF-904, B).Previous injury's rate ratio (2.60) had an interval that included no effect: not a shown HYROX risk factor.
Practice: pain under one station (knee in lunges, back on the sled) = remove that station's load first, not all training.
What the evidence does and does not show
Shows: Open-Division HYROX in recreational athletes is mostly running, and the aerobically fitter are faster (REF-510); in PRO/ELITE results running is ~50% of time and tracks rank (REF-903); PRO running is submaximal and fades (REF-902); HYROX injuries are mostly overuse, knee and lower leg (REF-904); interference is small and mostly scheduling (REF-330, REF-588); heavy strength helps endurance performance (REF-074); running after cycling is usually slower (REF-655).
Does not show: any HYROX training trial, the station-to-run penalty, what improves station time, a tested pacing strategy, HYROX fuel or taper, injury prevention, or women's pacing (the pacing study is men).
Evidence gaps
HYROX coverage is now race analyses and small observational studies (REF-510, REF-521, REF-901…REF-906); no HYROX training trial.
Station demands: Fernández-Navarrete et al. 2026, Ger J Exerc Sport Res — 10.1007/s12662-026-01151-3 (candidate, not an entry).
No entry for grip/carry endurance, high-repetition vs heavy work for station time, fluid in short indoor races, or resistance-to-run transitions.
How this entry was checked and changed
Why this level
(returned from B at verification 2026-10-01 — every Default and Decision row is practice or evidence-plus-practice; running-first rests on B+ HYROX race analyses (n=39696 Pro/Elite; n=186411 all categories) and small B observational studies; no HYROX training trial; A/A- meta-analyses on concurrent training and strength for endurance, all in non-HYROX populations; A-/B triathlon-transition entries used by analogy; the weekly layout, station drills, pacing and taper numbers are house practice)
Content correction 2026-09-27
Revision 2026-09-30: folded in REF-903 (Rappelt 2026, B+: running ~50% of PRO/ELITE time, most of the Top 100 gains; strength stations cost slower athletes more) and REF-901 (Fernández-Navarrete 2026, B+: percentile tables, values not quoted) into 'What decides the result'; the Pro Division row is no longer inference-only. Title, bold summary and Tier line no longer say the base has one HYROX study.
Revision 2026-09-30: folded in REF-902 (Gutiérrez-Hellín 2026, B, 24 PRO men: running at 84.13% of VT2 speed, fading from run 1 to run 8; Roxzone slowest) into race-day pacing; the pacing prescription stays practice (the study describes, it does not test a strategy).
Revision 2026-09-30: folded in REF-904 (Chittenden 2026, B, 89 athletes, 12 weeks) as an injury red flag: 14.4 injuries per 1,000 h, 73% overuse, knee and lower leg; previous injury's rate ratio 2.60 presented as not shown (CI includes no effect). REF-905 (B-) and REF-906 (C+) added as supporting reviews, not as grounds for the default.
Revision 2026-09-30: level raised B- → B — the running-first default now rests on B+ HYROX race analyses plus B studies, not one n=11 study; the weekly layout, station doses, pacing targets and taper remain house practice. Evidence-gap candidates that became entries removed; the Ger J Exerc Sport Res benchmarks paper stays a candidate.
Verification 2026-10-01: all 27 cited entries opened in the KB.md snapshot; every HYROX number (REF-510, 902, 903, 904, 521, 901) and every borrowed number (REF-074, 072, 330, 588, 640, 655, 656, 589, 539, 780, 619, 564, 010, 820, 533) matches the entry; no NUMBER, WRONG or per-ref LEVEL finding.
Verification 2026-10-01: transfer clauses are present where they are needed (triathlon cycle-to-run, CrossFit, rowing, trail, marathon, novice runners); inference and practice labels are attached to the week, station work, pacing, fuel, fluid and taper.
Verification 2026-10-01: no SOURCE flag. The evidence-gap candidate DOI 10.1007/s12662-026-01151-3 was looked up in Crossref: Fernández-Navarrete et al., 'Mapping the physiological demands of Hyrox: Which sections hit hardest?', Ger J Exerc Sport Res — real, and 'station demands' describes it correctly (the revision log's word 'benchmarks' for it is wrong; log not audited).
Verification 2026-10-01: cross-checked against SYNTH-44 (in-season 1 per 7-10 days; first-time lifter 2-4 weeks, practice), SYNTH-34 (two weeks, shorter for a small event = house reasoning), SYNTH-02 (<90 min 30-60 g/h; 90 min-3 h amateurs 60-90 g/h), SYNTH-42 (<8 weeks: specificity), SYNTH-45, SYNTH-21, SYNTH-31: consistent.
Verification 2026-10-01: REF-906 and REF-820 have no PMID but do have a DOI; REF-655 has a DOI and no PMID.
2026-10-01: label aligned with SYNTH-02: its <90 min row is consensus (REF-748), only the 90 min-3 h amateur cell is base practice.
2026-10-02: aligned with the rewritten SYNTH-42 (missing stimulus, one lever per block, level C+): "Specificity over the limiter" -> "Specificity, no new lever" (SYNTH-42's < ~8 weeks band).