SYNTH-67 · Guide
Supplements for endurance athletes
Fix the diet first, use caffeine for almost any race once a training trial shows it works for you, add a buffer or nitrate only when the event's duration matches what the evidence shows, keep creatine for strength and sprint goals, and skip the rest; loading schedules and timings beyond the position stands are practice
The question it answers
Which supplement, if any, for this athlete and event; dose, timing, expected effect; what to skip; what comes first.
Default (what to prescribe first)
1
Fix the diet before choosing a supplement. The IOC consensus says a full nutritional assessment should come before any supplement decision, and supplements make only a minor contribution to a nutrition programme (REF-358, A). The UCI statement treats energy availability (habitual 40-45 kcal/kg FFM/day) as the floor under everything else (REF-748, A; written for elite cyclists). The phrase "food first" is the base's own wording for these two rules.
2
Caffeine is the default race supplement. 3-6 mg/kg, most commonly ~60 min before; possibly effective from 2 mg/kg; ~9 mg/kg adds side-effects, not benefit; the most consistent gains are in aerobic endurance (REF-350, A, umbrella of 21 meta-analyses + ISSN stand). Across 21 time-trial studies the mean gain was 3.2 ± 4.3%, ranging from −0.3% to +17.3% (REF-083, A).
The UCI ticks caffeine for every cycling discipline (REF-748). Not for everyone: in a placebo-controlled 10-km cycling time trial in 101 competitive men, 4 mg/kg made the group 3% faster, but only CYP1A2 AA carriers gained (4.8% at 2 mg/kg, 6.8% at 4 mg/kg); AC carriers saw no effect and CC carriers were 13.7% slower at 4 mg/kg (REF-910, A-).
So the default stands only after item 4's training trial.
3
Add something else only if the event's duration fits (Decision table). The IOC names only caffeine, creatine, buffering agents and nitrate as having good evidence (REF-358); the UCI lists six with a case — caffeine, creatine, bicarbonate/citrate, β-alanine, nitrate, glycerol (AIS Group A) (REF-748).
4
Trial every one in training or a simulated race, never first in competition (REF-358); response varies with genetics, microbiome and habitual diet.
Decision — by event (order within a row is practice)
| Event / goal | Use | Skip or low priority | Source and how sure |
|---|---|---|---|
| Road race, TT over ~12 min, marathon, triathlon, gravel | Caffeine (after a training trial) | Nitrate for time-trial speed; buffers | Caffeine A (REF-350, REF-083); non-response and harm by genotype in one trial (REF-910, A-). Nitrate: time trial SMD −0.03 over 42 comparisons, no VO2max effect (REF-3493, A). The bicarbonate stand makes no claim beyond ~12 min (REF-354, A); β-alanine: duration moderated the effect, and the entry's own application calls it "not a priority" for long events (REF-351, A); beyond 25 min the ISSN stand calls it unresolved (REF-907, A) |
| Efforts of ~0.5-12 min: short TT, hill climb, pursuit, 1500 m | Caffeine; sodium bicarbonate; β-alanine (block, not race-week) | — | Bicarbonate: tasks of 30 s-12 min, single and repeated bouts, men and women (REF-354, A). β-alanine: in 0.5-10 min, capacity ES 0.50 but time-trial-type performance ES 0.11, CI crossing zero (REF-351, A) — inference: expect more in to-exhaustion than fixed-distance efforts |
| Criterium, XCO, cyclocross, interval-heavy blocks | Caffeine; bicarbonate; β-alanine | — | Bicarbonate helps multiple bouts (REF-354); intermittent vs continuous did not moderate β-alanine (REF-351). Inference: burst-racing fits the buffering window; UCI marks creatine and nitrate "more evidence needed" for MTB, cyclo-cross, road, gravel (REF-748) |
| Track sprint, BMX | Caffeine; creatine; nitrate | — | UCI ticks creatine and nitrate for BMX and track (REF-748); nitrate raised peak power (SMD 0.25) and muscular endurance (0.48) (REF-3493) |
| Strength block; women (masters: no specific entry) | Creatine monohydrate | Creatine before a power-to-weight goal (practice) | REF-352 (A for strength, B+ endurance; transfer from strength populations); REF-530 (B+, women) |
| Hot race | Caffeine 3-6 mg/kg; taurine as an n=1 trial, tested without caffeine (the entry carries an unchecked report that 5 mg/kg caffeine + taurine gave no benefit in heat) | — | REF-350; taurine 50 mg/kg 2 h before lengthened time to exhaustion 10% at 35 °C (REF-497, B, 11 healthy men) |
| Late in a long race, GI-sensitive | Caffeine mouth rinse (~5-s rinse, spat out) | As a replacement for swallowed caffeine | The use case is practice, not tested head to head. Pooled g = 0.12, most consistent for aerobic endurance and in the fed state (REF-422, A-) |
Supplement by supplement
Caffeine (REF-350, A; REF-249, A)
Not from a study; the base's own practice: start an untested athlete at 2-3 mg/kg; for races over ~2 h, 2-3 mg/kg before plus 2-3 mg/kg at 90-120 min (SYNTH-07).
Genotype: in the one trial, CC carriers were 13.7% slower after 4 mg/kg in a 10-km time trial while AA carriers gained (REF-910, A-; men only, subgroup sizes not reported); the base's earlier note is a C+ composite, and the ISSN calls genetic guidance preliminary (REF-249, Limitations; REF-446 is an unverified composite with no DOI, C+).A review of genetic testing for talent identification concludes such tests must be combined with other tools (REF-440, B+; talent identification, not caffeine — transfer). Practice: a with/without trial on matched sessions decides, genotype known or not; the REF-910 entry's inference is that a CC carrier who still wants caffeine tries 2 mg/kg in training.
Habituation. Ganio advises ≥7 days without caffeine for the full effect (REF-083); 28 days of 1.5-3 mg/kg/day removed the acute benefit in low-habitual users (REF-006, B, n = 18); Graham found users and non-users respond similarly (REF-116, B+); in 40 trained male cyclists 6 mg/kg improved a ~30-min time trial equally in low (58 mg/d), moderate (143 mg/d) and high (351 mg/d) habitual consumers (REF-909, B+).Inference: a new daily habit (REF-006) and an existing one (REF-909) are different questions; no entry tests pre-race withdrawal. Practice: habitual coffee drinkers do not withdraw before a race; save supplemental caffeine for key sessions and races.
Sodium bicarbonate (REF-354, A)
0.2 g/kg minimum, 0.3 g/kg optimal, 0.4-0.5 g/kg no extra benefit and more side-effects; 60-180 min before. Or 3-7 days at 0.4-0.5 g/kg/day split (0.1-0.2 g/kg at meals) to move GI effects off race day.
GI (bloating, nausea, vomiting, pain) varies between and within athletes; mitigate with a smaller dose, ~180-min lead, a high-carbohydrate meal or enteric-coated capsules. Part of the effect is placebo. Use before every training session may enhance adaptations.
Sodium citrate: listed with bicarbonate by the UCI (REF-748); not covered by the stand — practice only.
β-alanine (REF-351, A; 40 trials, n = 1461)
Overall ES 0.18; duration moderated the effect; training status (p = 0.559) and total amount ingested (p = 0.438) did not. Chronic loading is required — not a race-week supplement.
4-6 g daily for at least 2 to 4 weeks improves performance, most clearly in open end-point tasks or time trials of 1 to 4 min; four weeks of 4-6 g daily raises muscle carnosine; paraesthesia (tingling) is the only reported side effect, attenuated by divided doses of 1.6 g or a sustained-release product (REF-907, A, ISSN stand; some authors in the supplement industry). Practice: take it with meals and start at least 4 weeks before the target.
Nitrate / beetroot (REF-3493, A; REF-353, A)
Moves time to exhaustion (SMD 0.33), muscular endurance (0.48), distance covered (0.42), peak power (0.25) — not time trial or VO2max. Larger time-to-exhaustion effects with ≥6 mmol/day (372 mg/day) and >3 days of supplementation (REF-3493).
Who responds — contested. The umbrella review reports larger benefits in non-athletes and recreational athletes than in the highly trained (full text) (REF-3493); in McMahon's meta-analysis fitness level did not moderate the effect (REF-353). Most pooled reviews were AMSTAR-2 low to critically low, and products vary in nitrate content.
Not from a study in the base; the base's own practice: a concentrate a few hours before, after several days of loading; avoid antibacterial mouthwash (REF-353).
Nitrate lowers blood pressure (SBP −2.91, DBP −1.45 mmHg; REF-507, A): with antihypertensives, a physician first (SYNTH-20; the combination itself was not tested); with prescribed nitrates, a physician first as well — the base's own precaution, no entry covers them.
Creatine (REF-352, A strength / B+ endurance; REF-060, B+)
Raises muscle creatine, improves high-intensity performance and training adaptation, helps tolerate heavy training; safe up to 30 g/day for 5 years in the studies reviewed; no claim for prolonged aerobic performance (REF-352). 3-5 g/day or 0.1 g/kg/day (REF-060); either a loading phase then maintenance, or a low daily dose that raises stores more gradually over weeks (full text, REF-352).
When it makes sense (inference from these entries): strength blocks, sprint-decided events, repeated hard efforts, recovery from intense work (less damage and soreness, full text, REF-060). Women carry 70-80% lower endogenous stores; pre-menopause it appears to improve strength and performance, post-menopause 0.3 g/kg/day helps muscle and, with resistance training, bone (REF-530, B+).
Weight: expect some water-weight gain (REF-352, labelled practice); the base's practice is not to use it when power-to-weight or weight loss is the priority (REF-060). No entry quantifies the gain.
Don't bother / not supported
| Supplement | Verdict | Source |
|---|---|---|
| Exogenous ketones | Not recommended in any cycling discipline | REF-748 (A). Race-day case rests on "initial reports"; the one training trial (9 vs 9 fit men, twice-daily training) kept late-session power 15% higher, confounded by energy intake (REF-356, B); no effect on glycogen resynthesis (REF-054, B, n = 8) |
| High-dose vitamin C/E in training blocks | Do not mega-dose | Practice, not a study finding (SYNTH-70). In 54 young adults training mostly by running for 11 weeks, 1000 mg vitamin C + 235 mg vitamin E daily stopped the rise in muscle COX4 and PGC-1α seen on placebo, while VO2max rose 8% in both groups (REF-908, A-): blunted markers, no performance loss shown, and no gain. Tart cherry around one marathon sped isometric strength recovery (REF-009, B+, 20 recreational runners) — a short course, not daily |
| Daily polyphenol supplements (tea catechins, quercetin, curcumin, berry or soy extracts) through a strength block | No reliable added strength or muscle gain; no blunting shown either | Three-level meta-analysis of 11 small RCTs, 360 participants, none of them trained athletes: strength g = 0.25 (95% CI −0.01 to 0.51, low certainty), lean mass g = 0.13 (95% CI −0.07 to 0.33, low certainty), direct muscle morphology g = 0.05 (very low certainty); the authors draw no firm conclusion on benefit or lack of benefit (REF-1006, B+). It neither extends the vitamin C/E caution to polyphenols nor lifts it; endurance adaptation was not studied |
| Probiotics | Not a performance or race-day gut-symptom aid | The ISSN stand says specific strains can reduce upper respiratory tract infections and improve gut-barrier integrity, strain- and dose-dependent; its abstract names no strain or dose and claims nothing on endurance performance or gut symptoms during exercise (REF-982, A; narrative stand, many authors with industry ties). Symptoms: SYNTH-08 |
| BCAA against fatigue | No robust evidence either way | REF-481 (B+, 2006 narrative review; BCAA named only in its unchecked full text) |
| CoQ10 | Small, unstable chronic effect; none acute; certainty very low-low | REF-216 (A-, mostly Tier 1). In athletes: pooled VO₂max neutral (SMD = 0.03) and the oxidative-stress marker MDA not significant in the primary analysis (I² = 73.0%); authors: no robust ergogenic or physiological benefit (REF-955, B+, meta-analysis, 15 studies pooled) |
| Astaxanthin | No effect in 32 well-trained riders | REF-213 (B+) vs −1.2% in 12 recreational men (REF-211, B) |
| Vitamin D, omega-3 for performance | No performance evidence in the base; vitamin D only for a clinical reason | REF-355 (B+: high doses may be harmful); REF-357 (C+, mechanism only). Tension, on recovery only: one network meta-analysis ranks omega-3 first for "recovery outcomes" (SMD = 0.40 against placebo; REF-1008, C+) — that pool is creatine kinase, soreness scores and inflammatory markers rather than performance, several of its seven omega-3 trials were in older adults or measured no recovery outcome, and in the same paper omega-3 did not differ from placebo on strength or endurance, so the verdict in this row stands |
| Tyrosine, rhodiola | Unproven for endurance | REF-214 (C+, 12 mentally fatigued recreational cyclists; time to exhaustion +16% at 300 mg/kg, no rested condition); REF-215 (B+: exercise potential "seems minimal"); REF-022 (B+, footballers, intermittent-run test improved — transfer) |
Correcting a measured deficiency is a separate purpose (REF-358); e.g. 100-200 mg/day elemental iron when treating deficiency (REF-748), with a physician and a blood test.
Combinations
Bicarbonate + β-alanine gave the largest pooled effect, ES 0.43 (REF-351); the β-alanine stand says stacking may help only when β-alanine is 4-6 g daily for a minimum of 4 weeks (REF-907); the stand calls creatine or β-alanine with bicarbonate possibly additive, caffeine or nitrate with it unclear (REF-354).
Practice: add one at a time so the response can be attributed.
How to progress / periodise
Not from a study; the base's own practice: β-alanine and creatine start weeks out; nitrate for more than 3 days before (REF-3493); bicarbonate rehearsed in 2-3 hard sessions; caffeine rehearsed at race dose in a B-race.
When NOT to use it / red flags
Practice: no opaque blends or pre-workouts for tested athletes (contamination risk, REF-358).
Bicarbonate first used on race day; stacking untested products.
What the evidence does and does not show
Shows: group-level benefits of caffeine, bicarbonate, β-alanine, nitrate and creatine, each within the effort type named above.
Evidence gaps
β-alanine for endurance events beyond 25 min — unresolved by its stand (REF-907).
Whether antioxidant-blunted mitochondrial markers cost performance over blocks longer than 11 weeks (REF-908 showed no performance loss).
Chronic polyphenol supplements and adaptation in trained athletes, and with endurance training — REF-1006 pooled untrained and older adults doing resistance training only.
Creatine in masters endurance athletes; size of the creatine water-weight gain — no entry.
Sodium citrate, glycerol — no entries beyond the UCI list.
Supplement contamination rates — no verified entry.
How this entry was checked and changed
Why this level
(ranking rests on A-level umbrella reviews, meta-analyses and IOC/ISSN/UCI statements; β-alanine dosing on the ISSN stand; caffeine genotype on one A- RCT in 101 men; event mapping, loading schedules and caffeine cut-offs are practice)
Content correction 2026-09-27
Revision 2026-09-30: folded in REF-907 (Trexler 2015 ISSN β-alanine stand, A): the 4-6 g/day, 2-4 week, 1.6 g split-dose and paraesthesia line is no longer labelled practice (only 'with meals, start ≥4 weeks out' stays practice); its 'beyond 25 min unresolved' added to the long-event row and gaps; stacking condition added to Combinations.
Revision 2026-09-30: REF-910 (Guest 2018, A-, 101 competitive men) contradicts 'caffeine for almost any race' — CC carriers were 13.7% slower at 4 mg/kg, AC no effect. Minimal change: the title and the long-event row now say 'after a training trial'; Default 2 gains the genotype sentence; Genotype bullet rewritten on the trial (the C+ composite REF-446 kept as background). Caffeine stays the default race supplement.
Revision 2026-09-30: REF-909 (Gonçalves 2017, B+, 40 trained male cyclists) added to Habituation: equal gain in low/moderate/high habitual consumers; heading no longer 'contested'; practice line now 'habitual coffee drinkers do not withdraw before a race' (REF-909's practice line).
Revision 2026-09-30: REF-908 (Paulsen 2014, A-, n=54) replaces the 'not an entry' mention in the antioxidants row: markers blunted, VO2max +8% in both arms. Evidence gaps for the four candidates narrowed to what remains open. Level unchanged (B+).
2026-10-01: cross-reference repointed from a merged framework to the one that replaced it (SYNTH-01→31, 05→34, 09→70, 12/15→32, 19→26).
Revision 2026-10-01: the CoQ10 line in 'Don't bother / not supported' gains an athlete-specific source, REF-955 (Lu 2026, B+, systematic review and meta-analysis of randomised trials in athletes; graded B+ because each pooled outcome rests on five to seven small studies with I² = 73.0% on the primary outcome). It agrees with REF-216; verdict and level unchanged (B+).
Verification 2026-10-01: all 38 cited REF entries read in the KB.md snapshot; every dose, effect size, n and level next to a refId in Default, the decision table, the per-supplement sections and Combinations matches its entry except the items listed.
Verification 2026-10-01: Level B+ is honest — all five Default items rest on A/A- entries; event mapping, loading schedules and cut-offs are labelled practice in the Level line. No LEVEL-OVERALL.
Verification 2026-10-01: disclosed, not flagged — Ganio's 'at least 7 days without caffeine' (REF-083) versus the practice line 'habitual coffee drinkers do not withdraw'; the SYNTH states that no entry tests withdrawal.
Verification 2026-10-01: REF-497's '+10%' is the entry's figure; its own minutes (25.16 vs 22.43) give about 12%. Not a SYNTH error; the entry may need a check.
2026-10-01: SYNTH-20 was cited for 'prescribed nitrates', which neither it nor REF-507 covers. The citation now attaches to antihypertensives only; the prescribed-nitrates precaution stays, labelled the base's own with no entry behind it.
Revision 2026-10-01: SYNTH-67 had no probiotic line; one row added under 'Don't bother / not supported' from REF-982 (Jäger 2019 ISSN stand, A): what its abstract claims (strain- and dose-dependent URTI and gut-barrier effects) and does not (no strain or dose, nothing on endurance performance or exercise gut symptoms; industry ties). No default changed.
Revision 2026-10-01: the caffeine 'Sleep cost' bullet said the 6-h minimum was 'Drake 2013, not an entry'; Drake 2013 is now REF-973, so the bullet cites it and the '6-8 h' practice range is reduced to 'six hours is a floor (evidence), longer is practice' in line with SYNTH-07. To hold the length, the parenthetical 'the earlier SYNTH-09 is superseded' was dropped. Level B+ unchanged.
Revision 2026-10-02: added a row to "Don't bother / not supported" for chronic polyphenol supplements during strength training from REF-1006 (B+, three-level meta-analysis of 11 small RCTs, low to very low certainty): no reliable added strength or muscle gain and no blunting shown. No default changed, no level raised.
Revision 2026-10-02: added an evidence-gap line (polyphenols in trained athletes and with endurance training) and REF-1006 to the antioxidants cross-reference group.
Revision 2026-10-02: REF-1008 (Wang 2026, Nutrients, network meta-analysis of protein, creatine and omega-3; graded C+ after full-text appraisal — star network with no head-to-head trials, trial list breaks the review's own entry criteria, sample sizes differ from the source papers, one cited 'trial' is a meta-analysis) is recorded in the omega-3 row as ONE sentence of tension: it ranks omega-3 first for 'recovery outcomes' (CK, soreness, inflammatory markers), which is a different outcome from this row's 'for performance', and omega-3 did not differ from placebo on strength or endurance in the same paper.No default and no verdict changed; level unchanged (B+).
Revision 2026-10-02: nothing else from REF-1008 was folded. Its creatine-for-strength result agrees with the existing creatine row but adds no weight to A-level sources; its 'protein is most effective for endurance performance' claim did not survive appraisal (ranking only, not significantly different from the other two, pool mixes time to exhaustion, time trials and VO2max) and is not cited.