In athletes, dietary nitrate (beetroot juice or nitrate salts) against placebo showed no significant pooled effect on peak power (g 0.13, 95% CI −0.07 to 0.32), mean power (0.04), VO2max (−0.03), heart rate, lactate or RPE, in a meta-analysis of 45 randomised placebo-controlled trials with 659 athletes (41 crossovers; GRADE moderate for VO2max and heart rate, low for power); a peak-power gain in male-only studies did not differ from mixed-sex studies, and the overall peak-power estimate fell to about zero after trim-and-fill; time trials and time to exhaustion were not pooled. Read in full
Main caveat: The authors': risk of bias mostly "some concerns" (randomisation rarely described, 5 of 45 trials pre-registered); GRADE moderate to very low.
Xie and colleagues asked whether dietary nitrate, which helps recreationally active people in earlier reviews, helps athletes.
They pooled 45 randomised, placebo-controlled trials (41 crossovers) in 659 athletes of McKay Tier 2 or higher from cycling, running, triathlon, rowing, kayaking, skiing, swimming, team and combat sports and CrossFit, and analysed six outcomes: peak power, mean power, VO2max, heart rate, blood lactate and RPE.
Most trials gave one acute dose (29) and the rest loaded for 3-8 days, usually 6.2-8.4 mmol nitrate taken 2-3 h before testing. None of the six pooled effects was significant: peak power g 0.13, mean power 0.04, VO2max −0.03, heart rate −0.08, lactate 0.09 and RPE −0.17, all intervals crossing zero, with no detectable heterogeneity for the first four.
In subgroups, peak power rose in male-only studies (g 0.29) but the difference from mixed-sex studies was not significant, and trim-and-fill took the peak-power estimate to about zero; sport, exercise type, test, design and beetroot juice against nitrate salts made no significant difference.
Risk of bias was mostly "some concerns" because randomisation methods and registration were rarely reported. The authors conclude that nitrate gave no significant pooled benefit in athletes, that any benefit is small and context-dependent, and that the null results should not be read as proof of no effect.
Participants
659 athletes in 45 trials: 556 men, 88 women, 15 not reported; 36 trials men only, 6 mixed, 3 women only.
Sports: cyclists, triathletes, runners, rowers, kayakers, cross-country skiers, swimmers, soccer, ringette, hockey, fencing, taekwondo and CrossFit athletes.
Level: McKay Tier 2 (regional or provincial) or higher by the inclusion rule; the authors describe backgrounds from recreational to elite international level; no breakdown by tier, and no pooled age or VO2max is given in the text.
Excluded: non-athletes, and athletes with a musculoskeletal injury in the past 6 months.
Protocol
Search: five databases from inception; first run 6 May 2026, re-run 17 September 2026; reference lists and forward citation tracking (11 of the 45 trials came from tracking and monitoring); peer-reviewed articles only, no grey literature. Protocol registered on OSF on 30 June 2026, after the first search; the beetroot-versus-salts and crossover-versus-parallel subgroups were added after peer review.
Inclusion: randomised, placebo-controlled crossover or parallel trials; oral nitrate (beetroot juice, sodium or potassium nitrate) acute or for 2 days or more; placebo matched for look and taste (nitrate-depleted beetroot juice, blackcurrant juice, inert capsules); at least one of the six outcomes in a test with single bouts over 30 s.Excluded: co-supplements that could not be isolated, isometric or isokinetic strength, sprints under 10 s, cognition, resting blood pressure. A modified PEDro score of 16 of 20 or more was required for eligibility.
Supplementation in the trials: single acute dose in 29, loading of 3-8 days in 16 (median 6); 4.0 to 19.35 mmol a day, most often about 6.2-8.4 mmol (70-140 mL of concentrated beetroot juice); last dose usually 2-3 h before exercise (1.25-24 h); median crossover washout 7 days (4-14); most trials asked athletes to avoid antibacterial mouthwash.
Analysis: Hedges' g of pre-post change scores between nitrate and placebo (pre-post correlation 0.5 assumed); crossovers entered as if independent groups, which widens confidence intervals; one effect per study per outcome by a pre-set selection rule (dose nearest 6.2-8.4 mmol when several were tested); REML random effects in R (metafor, meta); subgroups when at least two levels had 3 or more studies; leave-one-out; sensitivity to pre-post and between-condition correlations of 0.3, 0.5 and 0.7; Egger's test and trim-and-fill.
VO2max outcome: peak oxygen uptake in a graded test; where a trial gave only submaximal values, the oxygen uptake at its top submaximal stage with complete steady-state data.
Risk of bias and certainty: RoB 2 by two reviewers, with three or more domains at "some concerns" counted as high risk overall; GRADE per outcome.
Funding and interests: no external funding; no conflicts declared.
Key Results
| Outcome (nitrate vs placebo) | Trials | Hedges' g (95% CI) | p | I² | GRADE |
|---|---|---|---|---|---|
| Peak power | 12 | 0.13 (−0.07 to 0.32) | 0.20 | 0% | low (risk of bias, publication bias) |
| Mean power | 17 | 0.04 (−0.13 to 0.21) | 0.65 | 0% | low (risk of bias, publication bias) |
| VO2max | 26 | −0.03 (−0.20 to 0.15) | 0.77 | 0% | moderate |
| Heart rate | 19 | −0.08 (−0.27 to 0.10) | 0.38 | 0% | moderate |
| Blood lactate | 10 | 0.09 (−0.30 to 0.48) | 0.64 | 59.79% | very low |
| RPE | 9 | −0.17 (−0.49 to 0.16) | 0.32 | 33.33% | very low |
Subgroups (none with a significant interaction):
| Moderator | Outcome | Result |
|---|---|---|
| Sex | Peak power | male-only g 0.29 (0.02 to 0.56), p = 0.04; mixed-sex −0.04 (−0.34 to 0.25); interaction p = 0.10 (Results) or 0.11 (Discussion, Conclusions) |
| Sex | VO2max; heart rate | women 0.09 (−0.37 to 0.54), men −0.05 (−0.24 to 0.15), p = 0.60; women −0.19 (−0.65 to 0.27), men −0.07 (−0.27 to 0.14), p = 0.62 |
| Source | VO2max | beetroot juice −0.02 (−0.21 to 0.16), nitrate salts −0.06 (−0.53 to 0.42), p = 0.90; not testable for the other outcomes |
| Sport | Mean power | cyclists −0.04 (−0.49 to 0.40), cyclists/triathletes 0.12 (−0.27 to 0.50), p = 0.60 |
| Sport | VO2max | cyclists −0.21 (−0.69 to 0.28), cyclists/triathletes −0.09 (−0.46 to 0.27), kayakers −0.08 (−0.64 to 0.47), runners −0.04 (−0.48 to 0.41), p = 0.96 |
| Exercise type | Peak power | endurance 0.00 (−0.24 to 0.25), strength/power 0.25 (−0.18 to 0.68), p = 0.32 |
| Exercise type | Mean power; heart rate | endurance 0.01, strength/power −0.05, p = 0.81; endurance −0.13, intermittent team sports 0.07, p = 0.41 |
| Test | VO2max | cycle ergometer test 0.20 (−0.22 to 0.63), submaximal cycling trial −0.31 (−0.80 to 0.18), p = 0.12 |
| Design | VO2max; heart rate | crossover −0.05, parallel 0.10, p = 0.56; crossover −0.03, parallel −0.57, p = 0.11 |
Publication bias: peak power, Egger significant and trim-and-fill imputed 5 studies, adjusted g −0.01 (−0.22 to 0.20); mean power, Egger significant (z = 2.33, p = 0.03), nothing imputed, estimate unchanged; trim-and-fill imputed 5, 0, 2 and 2 studies for VO2max, heart rate, lactate and RPE with estimates still not significant.
Sensitivity: no single study changed the direction or significance of any pooled effect; results held for assumed correlations of 0.3-0.7.
Risk of bias (RoB 2): sequence generation described in 9 of 45 trials; blinding low risk in 32, some concerns in 13; missing data low risk in 41; outcome measurement low risk in 42; selective reporting low risk in only 5 (pre-registered); overall 1 low, 33 some concerns, 11 high.
Practical Application
In trained athletes, nitrate did not raise peak power, mean power or VO2max on average. Across 45 placebo-controlled trials and 659 athletes every pooled effect sat near zero with intervals crossing it (peak power g 0.13, −0.07 to 0.32, low certainty; VO2max −0.03, moderate certainty); the authors stress that this is an imprecise null, not proof of no effect, and small benefits in some athletes or tests are not ruled out.
What this does to [SYNTH-67]: it fits the guide's "skip nitrate for time-trial speed" line and its VO2max line (REF-3493: no time-trial or VO2max effect), but it differs on peak power: the guide lists nitrate for track sprint and BMX with peak power up (SMD 0.25, REF-3493, pooled across reviews that include non-athletes), while in athletes alone this review found peak power not significant (g 0.13), the male-only gain not different from mixed-sex studies, and the estimate at about zero after trim-and-fill.Its peak-power tests were cycling or running efforts with bouts over 30 s, and sprints under 10 s were excluded, so it does not test the short sprints of the guide's track and BMX row directly. It did not pool time to exhaustion, the outcome the guide says nitrate moves, so it neither supports nor contradicts that line.
Inference, not tested here: for a trained endurance athlete, a beetroot shot before a race or test is a low-priority choice; anyone who wants to try it does so in training with the usual doses in these trials (about 6.2-8.4 mmol, 2-3 h before), knowing that the group effect in athletes was not detectable.
The source does not matter on the one outcome tested: beetroot juice and nitrate salts did not differ for VO2max (p = 0.90).
Limitations
The authors': risk of bias mostly "some concerns" (randomisation rarely described, 5 of 45 trials pre-registered); GRADE moderate to very low; few studies for subgroups, so interactions were underpowered; signs of small-study effects for peak and mean power; aggregate data only, so baseline fitness, dose per kilogram and plasma nitrite response could not be tested; crossovers analysed as independent groups, which widens intervals and may partly explain the null results; acute and chronic loading, doses of about 6 to 16 mmol and many test types pooled; registration after the first search, with two subgroup analyses added after review.
Outcomes chosen: time-trial time, time to exhaustion and economy were not pooled; time to exhaustion is where earlier reviews found nitrate effects (REF-3493), while time trials were null in that umbrella review and improved only in less-trained people in one earlier analysis cited here; the "VO2max" pool also took submaximal oxygen uptake when no maximal value was reported, so it mixes capacity with economy (our reading of the selection rule).
Population: "Tier 2 or higher" is the inclusion rule, yet the authors describe some backgrounds as recreational; 86% of participants were men (our arithmetic from 556 of 644 with a recorded sex).
Reporting inconsistencies: the sex interaction for peak power is p = 0.10 in the Results and 0.11 elsewhere; Egger's test for peak power is z = 3.03, p = 0.01 in the certainty section and z = 2.88, p = 0.02 in the publication-bias section; overall risk of bias is "1 low, 33 some concerns, 11 high" in the Results and "1 low and 44 some concerns" in the Limitations; the abstract places the trim-and-fill reduction to zero after the VO2max sentence, but in the text it belongs to peak power (for VO2max, 5 studies were imputed and the estimate stayed non-significant).
Funding and interests: no external funding; no conflicts declared.
Not read: supplementary tables (study characteristics, effect-size selection, risk-of-bias detail, sensitivity figures).
How this entry was checked and changed
Why this level
systematic review and meta-analysis of randomised trials whose risk of bias we read in the full text; held at A- because the risk of bias was serious (1 trial low, 33 some concerns, 11 high), the GRADE certainty is low for the power outcomes, and the trials are small (about 15 athletes each by our arithmetic)