REF-1377 · Yang et al., 2025 · BMC Sports Science, Medicine and Rehabilitation · Paper · PubMed · Free full text
Is HIIT better than sprint training for an athlete's VO2max?
In 51 trials, mostly team-sport athletes, HIIT, sprint and repeated-sprint training all beat usual training, with no clear difference between them (very low certainty). B+
Main caveat: The question it answers is not long versus short. Methods are HIIT, SIT and RST; HIIT spans 15-s to 300-s bouts and was never split by length.
Yang, Wang and Guan asked which of three interval methods raises VO2max most in athletes, and which protocol settings matter.
They pooled 51 studies (1,261 athletes) comparing high-intensity interval training (HIIT: below 100% VO2max or 80-95% HRmax, work bouts under 600 s), sprint interval training (SIT: all-out or supramaximal bouts usually 30 s or less with 2-4 min recoveries) and repeated-sprint training (RST: maximal 3-7-s sprints with up to 60-s recoveries) with each other, with continuous moderate training (CT) and with usual sport training (CON).
Against CON, the network estimates were RST g 1.04 (95% CI 0.76 to 1.32), HIIT 1.01 (0.76 to 1.25), SIT 0.69 (0.39 to 1.00) and CT 0.29 (−0.07 to 0.66); the P-score ranking was RST > HIIT > SIT > CT > CON, but the three interval methods did not differ significantly, and CINeMA rated every contrast between them very low certainty (HIIT v CON moderate, RST v CON high).
Within HIIT, a quadratic meta-regression across studies put the largest VO2max gain at work bouts of about 140 s and a work-to-recovery ratio of 0.85; the authors recommend "running-based HIIT (140 s work, 165 s recovery)" 3 times a week for 3-6 weeks, or SIT with recoveries under 97 s.
Athlete level did not moderate HIIT against usual training (Tier 2 g 0.93, Tier 3 0.87, Tier 4 0.96; p = 0.98).
Our reading: interval training of any of these kinds adds VO2max in sport athletes; the paper neither defines nor tests long versus short HIIT, its optimum near 2-2.5 min is a between-study curve, and its only 30/15 protocol (in adolescent ice-hockey players) was classed as SIT and gained more VO2max than the 4 × 5-min arm it was compared with, while in another small non-randomised trial in it 4-5-min reps gained more than 10-120-s reps at 95-105% VO2max (footballers, estimated VO2max).
Protocol
Design: PRISMA systematic review; registered (PROSPERO CRD42023435021); four databases searched on 1 April 2025; peer-reviewed English studies; two authors screened and extracted, a third checked.
Eligibility (Table 1): healthy athletes, no age limit; HIIT, RST or SIT under normal conditions; comparators CT (continuous moderate) or CON (usual specialised technical or tactical training without extra interval or continuous work); VO2max or VO2peak; "Randomized controlled, parallel and crossover trials". Excluded: non-athletes, injured athletes, combined interventions, altered conditions (hypoxia, heat, ergogenic aids), grey literature and non-controlled studies.
Definitions (Table 1): HIIT below 100% VO2max (80-95% HRmax), mean work bout under 600 s; RST maximal sprints (above 95% HRmax, above 100% VO2max) of 3-7 s with recoveries up to 60 s; SIT maximal or supramaximal efforts usually 30 s or less with 2-4 min of passive recovery.
How protocols were actually classed (Table S3): a 30/15 protocol (13 × 30 s "all out" / 15 s, three sets, cycling; Rønnestad 2021, adolescent ice-hockey players) and a 4 × 4-min protocol at 90-95% HRmax (Menz 2015) were entered as SIT; 10 × 30 s/30 s at 170% VO2peak (Mitranun 2018) and 15 s/15 s (Song 2024) as SIT; 20 × 15 s/15 s at 105-120% maximal aerobic speed (Ouerghi 2014) and 15-30 s/15-30 s at 110% maximal aerobic power (Paquette 2021) as HIIT.The HIIT category therefore spans 15-s to 300-s bouts.
n: n = 1261 athletes in 51 studies, 62 pairwise comparisons: HIIT v CON 13 studies (15 comparisons in Fig. 2A), SIT v CON 8 (9), RST v CON 14 (15), HIIT v CT 7 (8), SIT v CT 6 (7), HIIT v SIT 4, RST v HIIT 3, CON v CT 1; no direct RST v CT or RST v SIT.
Who (Table S2): football, basketball, volleyball, tennis, badminton, hockey, ice hockey, taekwondo, karate, grappling, canoe and kayak, swimming, alpine skiing and a few endurance groups; Tier 2-4; groups of 5-29 athletes.
What was matched: nothing across methods — not total work, work intensity, session duration or energy; each arm was compared with its own study's control or comparator, interval work was done on top of or within usual training.
Analysis: Hedges' g; random-effects frequentist network meta-analysis (netmeta, R 4.4.3) with P-scores; inconsistency by design-by-treatment and node splitting; three-level meta-analyses (metafor) for multiple arms per study; subgroups by athlete level, age, mode, duration, frequency, work-to-recovery ratio, team sport; meta-regression (REML) of work duration, work-to-recovery ratio, warm-up and recovery duration with baseline VO2max as covariate (and work duration as covariate for the ratio), models chosen by AICc among linear, quadratic, cubic and spline; Egger's test, funnel plots, trim and fill; CINeMA.
Quality: PEDro 4-7 (fair to good); random allocation scored 0 in 11 of 51 studies, concealed allocation met in 8, no blinding of participants, investigators or assessors (Table S4, our count). The limitations concede that non-randomised trials were included.
Funding and interests (as printed): Major Project of Philosophy and Social Science Research in Colleges and Universities of Jiangsu Province (2021SJZDA173) and Jiangsu Social Science Fund (24TYB008); the Crossref record also lists a 2025 project of the China Youth and Children Research Society (2025B41). "The authors declare no competing interests." No device maker, data platform or federation.
Key Results
| Comparison | Effect on VO2max, g (95% CI) | k | Certainty (CINeMA) |
|---|---|---|---|
| RST v CON (network) | 1.04 (0.76 to 1.32); P-score 0.88 | 15 direct | high |
| HIIT v CON (network) | 1.01 (0.76 to 1.25); P-score 0.85 | 15 direct | moderate |
| SIT v CON (network) | 0.69 (0.39 to 1.00); P-score 0.51 | 9 direct | low |
| CT v CON (network) | 0.29 (−0.07 to 0.66), p = 0.11; P-score 0.23 | 1 direct | very low |
| HIIT v SIT (direct, three-level) | 0.39 (−1.33 to 2.11), p = 0.52 | 4 | very low |
| HIIT v RST (direct, three-level) | 0.16 (−0.86 to 1.17), p = 0.58 | 3 | very low |
| HIIT v CT (three-level) | 0.69 (0.09 to 1.29); prediction interval −0.62 to 2.00 | 8 | low |
| SIT v CT (three-level) | 0.39 (0.01 to 0.77) | 7 | moderate |
Network heterogeneity and consistency: τ² = 0.13, I² = 36%; between-design inconsistency Q = 3.05, p = 0.88; no direct-indirect disagreement (Table S6); Egger p = 0.42 for the network.
Within HIIT, work-bout length (between-study meta-regression, Table S13 and Fig. 7B): quadratic model, linear and squared terms each p = 0.01, baseline VO2max not associated (p = 0.94); peak at about 140 s. Work-to-recovery ratio (spline, Table S14): peak at 0.85, adjusted for work duration.
Work-to-recovery subgroups, HIIT v CON (Fig. 4, Table S9): ratio above 1 (mean work 240 s, recovery 174 s) g 1.02 (0.40 to 1.64), k 5; 0.5-1 (117.9 s, 156.4 s) 0.79 (0.20 to 1.38), k 7; below 0.5 (30 s, 170 s) 0.93 (−0.22 to 2.09), k 3; between-subgroup p = 0.81.
Training status: HIIT v CON did not differ by tier (Tier 2 0.93, Tier 3 0.87, Tier 4 0.96; p = 0.98). HIIT v CT did (p < 0.01), driven by one Tier 2 study (1.78; Tier 3 0.32, Tier 4 0.75).
The 30/15 protocol (Table S3, our reading): in Rønnestad 2021 (adolescent ice-hockey players, Tier 3, 9 weeks), the 30/15 group, classed SIT, went from 60.5 to 62.5 mL/kg/min and the 4 × 5-min group, classed HIIT, from 60.8 to 60.2; it is one of the 4 HIIT v SIT comparisons.Another, Pierros 2023 (footballers, Tier 2, 4 weeks, VO2max estimated from the Yo-Yo test), went the other way: 4-6 × 240-300 s at 85-100% VO2max (HIIT, 16 players) from 54.88 to 59.74 and 3-10 × 10-120 s at 95-105% VO2max with 20-120-s recoveries (classed SIT, 13 players) from 56.5 to 55.33; neither trial scored for random allocation (Table S4).
Authors' conclusion: "RST, HIIT, and SIT all enhance VO2max in athletes, with RST demonstrating the strongest probabilistic efficacy. Optimal protocols include 3–6 weeks of running-based HIIT (140 s work, 165 s recovery) or SIT (≤ 30 s sprints, < 97 s recovery) at 3 sessions per week."
Numbers that do not add up: HIIT v CON is "n = 13 studies" with 14 references cited and 15 comparisons in Fig. 2A; the SIT recovery threshold is 97 s in the abstract, results and conclusion and 96 s in the discussion and Fig. 7D; the cycling subgroup in Fig. 4 prints 0.34 (0.54, 1.23), Table S9 0.34 (−0.54, 1.23); Table S5 is titled P-score and SUCRA but gives P-scores only.
Mechanisms
Measured: all three interval methods raised VO2max against usual sport training by similar amounts; within HIIT, gains peaked at mid-length work bouts (about 140 s) across studies.
Authors' interpretation: recovery intervals must allow phosphocreatine resynthesis and acid-base recovery so that the intensity of the next bout holds; HIIT of 2-3 min with a work-to-recovery ratio near 1 stimulates lactate oxidation and VO2max; SIT with recoveries over about 97 s has too little aerobic demand.
Inference: a quadratic across studies of different sports, lengths and intensities finds where the studied protocols happened to work best, not what a shorter or longer bout would do for one athlete; the fitted curve at 30-s work bouts is still above zero (Fig. 7B, our reading of the point estimate).
Inference: in sport athletes whose usual training holds little high-intensity aerobic work, almost any added interval work raises VO2max; differences between methods are hard to see against that.
Practical Application
Is HIIT better than sprint training for an athlete's VO2max? In 51 trials, mostly team-sport athletes, HIIT, sprint and repeated-sprint training all beat usual training, with no clear difference between them (very low certainty). B+
Any interval method adds VO2max to sport training: g about 0.7-1.0 against usual training at Tier 2-4, with no difference by tier for HIIT (B+, mostly team-sport athletes, mostly running).
The 140-s "optimum" is a guide, not a rule: it comes from a curve across studies (B+); the authors' prescription is running-based 140 s on, 165 s off, 3 sessions a week for 3-6 weeks.It is weak, indirect evidence on bout length and none on 30/15 as such: the curve's short end is mostly 30-s bouts with 1-4-min recoveries (Table S3, our reading), and of the network's two small non-randomised head-to-heads of long reps against shorter non-sprint work, the 30/15 protocol (classed as sprint training) gained more VO2max than 4 × 5 min while 4-5-min reps gained more than 10-120-s reps at 95-105% VO2max (footballers, estimated VO2max).
Sprint intervals: keep recoveries short: SIT gains were no longer significant when recoveries exceeded about 97 s (between-study meta-regression).
Not from this paper; the base's own practice: for trained cyclists the format choice stays with SYNTH-33; this paper is about sport athletes and does not compare 30/15 with long reps.
Limitations
The question it answers is not long versus short. Methods are HIIT, SIT and RST; HIIT spans 15-s to 300-s bouts and was never split by length; the work-bout curve is a between-study meta-regression without the number of points reported.
Misclassified protocols. A 4 × 4-min protocol at 90-95% HRmax (Menz 2015, a fit for the paper's HIIT definition) and 10-120-s reps at 95-105% VO2max (Pierros 2023, not the all-out sprints the SIT definition names) entered the network as SIT; so did a 30/15 protocol with 15-s recoveries (Ronnestad 2021, recorded as all-out, against a SIT recovery of usually 2-4 min), while the same trial's 4 × 5-min arm, also recorded as all-out, was classed HIIT.
Certainty. Every head-to-head contrast between interval methods is very low certainty (CINeMA); subgroups often rest on 1-3 studies.
Risk of bias. No blinding, rare allocation concealment, 11 studies without random allocation (our count from Table S4) despite an eligibility table that asks for randomised trials.
Who. Mostly team-sport and other non-endurance athletes, mostly men, often adolescents; HIIT against usual training was nearly all running; usual training differs by sport.
Reporting. Study counts, the SIT recovery threshold and one confidence interval differ between text, figures and tables (see Key Results).
Interests. Public research funding; no competing interests declared.
How this entry was checked and changed
Why this level
a registered network meta-analysis of mostly small trials would sit at A- (PEDro 4-7, no blinding, concealed allocation met in 8 of 51, 11 of 51 studies scored 0 for random allocation although the eligibility table asks for randomised trials); held one step lower because the paper's question — which interval method is best — rests on very-low-certainty contrasts by its own CINeMA rating, the 140-s optimum is a between-study meta-regression, and protocols were placed in categories that contradict the paper's own definitions (a 4 × 4-min protocol and a 30/15 protocol entered as sprint training). Read from the full text, both supplements and the figures (open access, CC BY-NC-ND 4.0)
Full text (checked 2026-10-28)
51 eligible studies (1,261 athletes); search conducted on April 1, 2025; PROSPERO CRD42023435021 (full text)
HIIT intervention, involving high intensity (< 100%VO2max, 80-95%HRmax), with a mean work duration of < 600 s (full text)
RST (NMA's g = 1.04, 95% CI [0.76, 1.32), HIIT (NMA's g = 1.01, 95% CI [0.76, 1.25), SIT (NMA's g = 0.69, 95% CI [0.39, 1.00]); CT 0.29, 95% CI [-0.07, 0.66], p = 0.11 (full text)
Random effects model τ² = 0.13, I² = 36%; between-design inconsistency Q = 3.05, p = 0.88 (full text)
HIIT compared to SIT (g = 0.39, 95% CI [-1.33, 2.11], p = 0.52); HIIT compared to RST (g = 0.16, 95% CI [-0.86, 1.17], p = 0.58) (full text)
The highest increase in work duration per interval was observed at 140 s, achieving optimal benefits at a work-to-recovery ratio of 0.85 (full text)
Fig. 4, HIIT v CON by athlete level: Tier 2 0.93, Tier 3 0.87, Tier 4 0.96, p 0.98, transcribed from the figure (full text)
Ronnestad et al., 2021: SIT, 13 reps, 30s, 15sA, VO2max 60.5±4.8 to 62.5±2.6; HIIT, 4 reps, 300s, 150sA, 60.8±4.7 to 60.2±4.3 (full text)
Fig. 2E, CINeMA: HIIT:CON moderate, RST:CON high, SIT:CON low, HIIT:SIT and HIIT:RST very low, transcribed from the figure (full text)
Funded by the Jiangsu Social Science Fund (24TYB008); the authors declare no competing interests (full text)