SYNTH-57 · Guide
Preparing for a 70.3 or Ironman
Keep most hours easy, grow one long ride and one long run, ride the bike for the run, and train the gut for the carbohydrate the race needs; the training evidence is 9 and 4 triathletes, so the hours are practice
The question it answers
An athlete has a 70.3 or a full Ironman: hours per sport and how hard, how far the long ride and long run grow, bricks, bike pacing that leaves a run, carbohydrate per hour and gut training, then sodium, heat, taper, strength and masters. Shared short-course material (swim, open water, transitions, sessions per sport, brick effect) is in SYNTH-52.
What decides the result
Energy supply. Faster athletes burn energy faster and need more carbohydrate per hour, while a slower race runs at a lower rate for a similar total (modelled: 20.1, 16.6 and 13.3 kcal/min for ~8, ~9 and ~13 h finishes; 9626, 9062 and 10,425 kcal). Elite finishes are about 8 h and amateur finishes 14-15 h (REF-142, C+, opinion piece; energy costs modelled from race times and published economies, not measured).
Pacing that fades or swings. Six well-trained men on a flat Ironman course faded from 239 to 203 W and from 89 to 82 rpm lap by lap (REF-786, B, observational, one race). At Kona, the athletes who did better against their goal kept moving on descents and kept their heart rate steadier from climb to descent (REF-659, B, observational).In 12 well-trained men, a 1-h ride swinging 40-140% MAP slowed the following 9.3-km run 42 ± 37 s against the same mean power held steady (REF-831, B, a short-course lab protocol — transfer to long course is inference).
A controlled swim. In 9 male Ironman triathletes the swim was raced at the day's highest HR (149 vs 137 bike, 136 run bpm), bike and run HR sat below both ventilatory thresholds, and a larger swim-bike HR gap went with slower later stages (REF-836, B, one race, correlational).
Inference: which leg decides age-group long-course placing is not in the base; SYNTH-52's "the run decides" is elite draft-legal. In elite long-course men, peak cycling power tracked their bike split in a short-distance triathlon (r = −0.83; subgroup size not given), and their lab physiology matched short-course specialists' except for a slower swim (REF-838, B, 15 men).
Default (what to prescribe first)
For a healthy Tier 1-3 age-grouper with at least ~20 weeks to a 70.3 or ~30 weeks to an Ironman (not from a study; the base's own practice):
Mostly easy, and keep the middle out. Nine triathletes were monitored by heart rate through one 18-week Ironman build (sex and level not in the abstract). Those who trained more in total and more below the aerobic threshold raced faster (r = −.69 and −.92).A larger share of time between the thresholds went with a slower race (r = .94), although 59% of the race itself was spent in that zone (REF-784, B). The data are correlational and total hours and zone-1 hours rose together, so part of the "zone 1" effect may be volume (that entry's inference). How easy is easy → SYNTH-51.
Hours. Not from a study; the base's own practice: about 7-12 h a week at peak for a 70.3 and 10-16 h for an Ironman. Roughly half goes to the bike, a quarter to a third to the run, and the rest to the swim. Below those ranges (also practice), finishing comfortably is a realistic goal; a time goal is not.
Do not cut the swim to feed the bike. In four elite triathletes modelled over 40 weeks, cycling training carried over to running performance on a submaximal run test (r = 0.56). Running training drove race results (r = 0.52). Swimming neither gained from nor gave to the other two sports (REF-785, B, elite case series; the swim model fitted poorly, r = 0.37).That model had no cycling performance test, so it could not detect run → bike transfer. REF-537 (B+, narrative review) says that direction is likely the larger one. Inference: the two entries do not contradict each other.
Heavy strength twice a week in the base phase. In 15 triathletes over 14 weeks, it improved running economy against endurance-only training, with VO2max unchanged (REF-082, B; strength ran alongside endurance for all 14 weeks and no phase was tested — "base phase" is practice). Dose → SYNTH-44 and SYNTH-43; placement around long sessions → SYNTH-31.
Decision
| Situation | Do this | Source and how sure |
|---|---|---|
| 70.3 amateur, ~5-7 h finish (base estimate) | Work toward the 3-6 h amateur cell of the matrix (80-100 g/h; on the run expect the low end or less) only as far as rehearsal allows; a finish beyond 6 h starts at 60-80 g/h, set by rehearsal (SYNTH-02) | SYNTH-02 B: the 3-6 h amateur cell is 80-100 g/h (practice). REF-787 B+: the mean 70.3 intake was 65 g/h |
| Ironman amateur, >10 h | Set the rate by rehearsal, usually below the 3-6 h cell. Do not drop carbohydrate | SYNTH-02: ">6 h start lower and are set by rehearsal". REF-787: in both Ironman races, more total carbohydrate went with a faster finish (r = −0.55, −0.48) and more nausea (correlational) |
| Slow amateur asks about low-carb or keto | Do not prescribe it; keep a carbohydrate plan | REF-142 C+ (opinion, modelled) says a low-carbohydrate or ketogenic diet "might" suit some finishers beyond 12 h, mainly those with gut trouble on race-day carbohydrate who aim simply to finish; performance effects "not known". REF-663 B+ (ultrarunners) says LCHF is not supported for performance. The resolution is the base's own |
| History of GI trouble | Start lower, ramp more slowly, and test every product in training (the base's own practice) | REF-787: symptom scores tracked GI history (r = 0.37 upper, 0.51 lower). About 31% of Ironman athletes had serious symptoms |
| Running volume limited (injury, impact tolerance) | Hold running frequency and add bike hours | REF-785 cycling → running transfer (elite, n=4). Substitution is that entry's inference, not tested. REF-623 B+: keep weekly run increases under 20% (half-marathon runners, a transfer) |
| Hilly course | Keep working at a controlled effort on descents, and avoid large HR swings from climb to descent | REF-659 B, Kona, observational. Rehearsing descents is that entry's house practice |
| Hot race | 8-10 heat exposures beforehand. Expect less of the drink to be used, and pace by RPE/HR | SYNTH-04 B+ (the dose is practice built on a pooled literature). REF-461 B: 10 men, 32 vs 19 °C, exogenous glucose oxidation −15%. REF-525 A-: heat worsens GI markers |
| Masters | Keep the intensity, and widen the gaps on what recovery shows, not on age. Keep strength 2×/week | SYNTH-35 B. SYNTH-44 masters row |
How to progress / periodise
Horizon. More than ~16 weeks out, work the missing stimulus in the leg that costs the most time. Inside ~8-12 weeks, make training race-specific (SYNTH-42; the bands are house defaults).
Long ride. Not from a study; the base's own practice: grow it by 15-30 min every week or two. Peak at ~3-3.5 h for a 70.3 and ~5-6 h for an Ironman, with the last one about 3 weeks out. Ride it at the rate of fuelling planned for the race. In REF-181 carbohydrate protected VT1 power over 150 min, while the 5-min time trial after it did not differ significantly between conditions.
Long run. 26 well-trained male runners were matched on 10-km time. Economy drifted less over 90 min in those doing regular ≥90-min long runs (+3.1% vs +6.0%), and the drift correlated with the length of the weekly longest run (r = −0.67) (REF-160, B, cross-sectional).Not from a study; the base's own practice: cap it near 2-2.5 h for an Ironman and ~1.5-2 h for a 70.3, because the time on the feet is what injures. Keep weekly run distance increases under 20% (REF-623).
Gut training. The gut is trainable, with better gastric emptying and more intestinal carbohydrate transporters, but no timeline is established (REF-008, B+, review). Not from a study; the base's own practice (SYNTH-02): add 15-20 g/h a week from the current intake.Above ~60 g/h use glucose + fructose. REF-3490 (B+, perspective review) proposes 120 g/h as a trained-athlete ceiling, argued from oxidation, not from finishing time; past 6 h that ceiling does not apply (SYNTH-02) — reported intakes in races of ~12-24 h are 30-66 g/h, elite case studies 80-100 g/h (REF-3490, full text) (corrected 2026-10-12).
Taper → [SYNTH-34]. Across 27 studies the best taper lasted 2 weeks, cut volume 41-60% and kept intensity and frequency (REF-107, A, mixed sports). In a 14-study meta-analysis, taper bands up to 21 days worked and the ≥22-day band was not significant (REF-3492, A). SYNTH-34 allows longer only for very long events (house reasoning).Inference: an Ironman taper is 2-3 weeks, never a month. If carbohydrate-loading, depletion days are unnecessary: a 50%-then-70% regimen stored as much glycogen as the depletion diet, though neither sped a 20.9-km run and longer races were untested (REF-774, B, trained runners).
Race pacing (bike). Plan the first lap so the fade stays small, ride as evenly as the course allows (REF-831, a 1-h short-course protocol — a transfer); do not bank time. Riders in REF-786 held the same average power into a headwind as with the tailwind. The optimal pacing for events over 4 h is not established (REF-010, B+, 2008 review).A narrative review argues the anaerobic threshold is too hard for an Ironman bike, as lab time to exhaustion there peaks at 255 min, with fuel and fluid limits on top (REF-833, C+, no tested target); pacing research is mostly Olympic-distance (REF-837, C+).Not from a study; the base's own practice: an Ironman bike at ~65-75% of FTP and a 70.3 at ~75-82%, confirmed on a long-ride brick. The run pace comes from rehearsals, never from fresh-run pace.
Fluid and sodium. Drink to thirst, not to a schedule; in a hot race plan bottles from a weighed sweat rate, never above the measured loss, and let thirst override the plan downward (SYNTH-68).The international consensus attributes exercise-associated hyponatraemia (EAH), and all known deaths from it, to drinking more than is lost, and says sodium in drinks cannot prevent EAH in someone who keeps overdrinking (REF-776, A-, consensus over mostly observational data).Finishing heavier than at the start, or losing no weight over a long race, is its warning sign (REF-776). Plan sodium against the fluid plan, not against sweat losses alone (REF-220, B+, narrative review).
When NOT to use it / red flags
Not from a study; the base's own practice, built on REF-776's treatment section: a worsening headache, nausea or vomiting, confusion or unusual drowsiness after drinking a lot, during or after a long event, means medical help now and no more water or sports drink until assessed. Never advise drinking more to fix them.
Ibuprofen during the race: in 89 ultramarathoners over 80 km, 400 mg every 4 h was followed by acute kidney injury in 52% vs 34% on placebo (number needed to harm 5.5; the CI crossed zero) (REF-892, A-, desert running — a transfer). Not from a study; the base's own practice: no NSAIDs during a long-course race.
Open water, overreaching and running-injury red flags → SYNTH-52.
Race-intensity work is not the bulk of training (REF-784, with the caveats above).
What the evidence does and does not show
Shows: more easy volume went with a faster Ironman (n=9, correlational); no transfer into swimming (n=4 elites); riders fade over 180 km; more carbohydrate went with faster finishes and GI history with symptoms (self-report survey); the gut is trainable; taper shape; overdrinking causes EAH (consensus).
Does not show: an hours split, long-session lengths, brick dose or power targets that beat alternatives. Whether a more even Ironman bike yields a faster marathon (REF-786 has no run data; REF-831 is a 1-h ride and a 9.3-km run). Whether eating more causes a faster finish (REF-787 cannot separate that from experience). Anything 70.3-specific beyond REF-787's intake row.
Evidence gaps
How this entry was checked and changed
Why this level
(intensity distribution rests on one correlational season, n=9 (REF-784); cross-sport transfer on an n=4 elite model (REF-785); bike pacing on two small observational race studies; fuelling on one B+ field survey plus B+ reviews. Hours, long-session progression, brick dose and race targets are house practice)
Content correction 2026-09-27
Revision 2026-09-29: folded in REF-831 (Etxebarria 2013, B: a surging ride cost the following run at the same mean power) and REF-836 (Barrero 2014, B: Ironman swim raced hardest, a bigger swim-bike HR gap went with slower later stages) into 'what decides the result' and bike pacing ('ride as evenly as the course allows' now has a study, by transfer); REF-838 (Millet 2003, B), REF-833 (Laursen & Rhodes 2001, C+) and REF-837 (Wu 2014, C+) into what decides and bike pacing; REF-892 (Lipman 2017, A-: ibuprofen in an 80-km ultra, AKI 52% vs 34%) as a new red flag with a labelled practice rule; REF-832 (Gosling 2008, C+) in gaps.House power targets unchanged. Gaps narrowed; level unchanged (B-).
Verification 2026-10-01: every refId outside Cross-references and the revision log was opened in the KB.md snapshot (31 entries). All numbers match the entries (r = −.69/−.92/.94, 59%; r = 0.56/0.52/0.37; 239→203 W, 89→82 rpm; 42 ± 37 s; 149/137/136 bpm; −3%/−6%; +3.1%/+6.0%, r = −0.67; 65 g/h; r = −0.55/−0.48; r = 0.37/0.51; ~31%; 52% vs 34%, NNH 5.5; 255 min; 41-60%, 2 weeks, 21 days), as do all level tags and the levels quoted for SYNTH-02/04/32/35.
Verification 2026-10-01: no WRONG and no NUMBER findings. Hours, long-ride and long-run lengths, brick dose and %FTP targets are all labelled practice in the same sentence.
Verification 2026-10-01: Level B- is acceptable: the card states that the training default rests on n=9 correlational and n=4 modelled data and that the hours are practice; the fuelling, taper and hyponatraemia parts rest on B+ to A sources.
Verification 2026-10-01: Decision row 2 quotes SYNTH-02 as ">6 h start lower and are set by rehearsal"; SYNTH-02's cell reads "Start lower, set by rehearsal: begin at 60-80 g/h" — same meaning, not verbatim.
2026-10-01: the 70.3 row's own '60-90 g/h' disagreed with SYNTH-02's cells for a 5-7 h day; replaced with a pointer and SYNTH-02's current wording (3-6 h amateur 80-100 g/h, runners the low end or less; beyond 6 h start at 60-80 g/h). Fluid line now carries SYNTH-68's hot-race rule (a plan from a weighed sweat rate, never above the measured loss).
2026-10-02: aligned with the rewritten SYNTH-42 (missing stimulus, one lever per block, level C+): "work the limiter of the leg" -> "work the missing stimulus in the leg"; cross-reference label.
Revision 2026-10-14: REF-142 claims re-checked against the full text (Current Opinion with a modelled energy budget, regraded B+ -> C+; "a slower athlete works where fat covers more of the demand" -> lower modelled rate, similar total cost (20.1/16.6/13.3 kcal/min; 9626/9062/10,425 kcal); "no race data" -> costs modelled from race times and published economies; the low-carbohydrate row now carries the paper's conditions and its "not known" on performance).Decision rows unchanged.