Do two interval sessions a week beat one?
In 41 active young men doing 4×4-min runs for 6 weeks, VO2max rose ~1% at one a week, ~9-11% at two or three; the between-arm test was not significant. B+, transfer.
Main caveat: The frequency effect on VO2max is not statistically shown. The group-by-time interaction was p = 0.230 (time to exhaustion p = 0.143).
Zhang et al. randomised 45 physically active male students to the same 4×4-min treadmill interval session once, twice or three times a week for 6 weeks. Intensity started at at least 85% of the speed at VO2max (85-90% HRmax) and rose to at least 95% of that speed (90-95% HRmax) by the last week, with 4 min of walking between reps.
Four men withdrew because they could not tolerate the load: none at one session a week, one at two, three at three (completion 100%, 93% and 80%). Among the 41 who finished, VO2max rose from 49.78 to 50.46 mL/kg/min at one session a week, from 49.04 to 53.58 at two and from 47.90 to 53.09 at three (about +1%, +9% and +11%, our arithmetic).
The main effect of time was significant, but the group-by-time interaction was not (p = 0.230), and the paper reports no test of each group's own change; the "marked increase at two and three a week" in the abstract is a reading of the means. Time to exhaustion rose in all groups (interaction p = 0.143).
Blood lactate after 8 min at 9 km/h was the one outcome where frequency clearly mattered: it fell from 2.26 to 1.77 and from 2.11 to 1.71 mmol/L at two and three a week against 2.23 to 2.04 at one (interaction p = 0.014; both lower than one a week after training). Oxygen cost at 9 km/h, the ventilatory threshold and RER did not change.
The authors conclude that two a week gives the best balance of benefit and tolerability.
Our reading: consistent with REF-1291 (the same session at the same three frequencies in recreational runners, also non-significant between arms), it says that in active young men a third weekly 4×4 added little over a second and cost retention, and that one a week barely moved VO2max in 6 weeks; it is not a test in endurance athletes and not proof that two beats one.
Protocol
Design: randomised parallel trial, three arms, 6 weeks; computer-generated sequence (Matlab), allocation concealed in sealed opaque envelopes opened after baseline tests; assessors not blinded; analysis of completers only. Ethics approval Capital University of Physical Education and Sports (No. 2025A168).Registered retrospectively at the Chinese Clinical Trial Registry on 2 December 2025 (ChiCTR2500113733). A power analysis (effect size 0.3, power 0.80) asked for at least 24 participants.
Participants: n = 45 men recruited from one sport university (15 per arm), aged 19.93-20.33 y (arm means), body mass 73.60-76.00 kg; regular structured exercise at least 3 times a week (60 min or more) for at least 3 years, in resistance training, badminton, basketball or soccer; no competitive sport in the past 12 months. No women and no sedentary people.
Arms: one session a week (Monday), two (Monday, Wednesday) or three (Monday, Wednesday, Friday).
Session: 4 × 4 min treadmill running, 4 min walking at 4 km/h between reps. Weeks 1-2: at least 85% of pre-test vVO2max, HR 85-90% HRmax; intensity raised each week; final week: at least 95% of vVO2max, HR 90-95% HRmax. Polar HR monitor at every session; sessions supervised.
Tests (before and after): incremental treadmill test (5 min at 8 km/h, then +1 km/h each minute, 1% incline) with breath-by-breath gas analysis (Cortex META-LYZER 3B) for VO2max and time to exhaustion; 72 h later, 8 min at 9 km/h for VO2, HR, ventilatory threshold, RER and earlobe lactate (Lactate Pro 2).
Analysis: 2 (time) × 3 (group) repeated-measures ANOVA, Bonferroni post hoc tests, partial eta squared; significance p < 0.05. Withdrawals (4) excluded; 41 analysed (15, 14 and 12).
Funding and interests: the authors declare that no financial support was received and no conflicts of interest.
Key Results
| Outcome (mean ± SD) | 1 a week (n = 15) | 2 a week (n = 14) | 3 a week (n = 12) | Group × time |
|---|---|---|---|---|
| VO2max, mL/kg/min, pre → post | 49.78 ± 4.03 → 50.46 ± 4.24 | 49.04 ± 4.71 → 53.58 ± 5.58 | 47.90 ± 4.40 → 53.09 ± 6.97 | F = 1.527, p = 0.230, η² = 0.074 |
| VO2max change (our arithmetic) | +0.68 (+1.4%) | +4.54 (+9.3%) | +5.19 (+10.8%) | — |
| Time to exhaustion, s | 606.20 ± 43.36 → 638.73 ± 49.08 | 600.50 ± 39.63 → 669.64 ± 36.55 | 612.91 ± 28.58 → 673.42 ± 30.16 | p = 0.143 |
| HR at 9 km/h, beats/min | 141.87 ± 10.94 → 137.20 ± 8.55 | 141.07 ± 8.72 → 133.35 ± 6.08 | 139.08 ± 7.22 → 131.83 ± 6.18 | p = 0.581 |
| Lactate after 8 min at 9 km/h, mmol/L | 2.23 ± 0.27 → 2.04 ± 0.18 | 2.26 ± 0.21 → 1.77 ± 0.14 | 2.11 ± 0.17 → 1.71 ± 0.12 | p = 0.014, η² = 0.200 |
| VO2 at 9 km/h, mL/kg/min | 30.56 ± 3.82 → 28.92 ± 3.47 | 29.78 ± 2.60 → 27.94 ± 1.77 | 30.35 ± 3.60 → 28.82 ± 3.36 | p = 0.989 |
| Completed the 6 weeks | 15 of 15 (100%) | 14 of 15 (93%) | 12 of 15 (80%) | — |
Between arms (the comparison that tests frequency): not significant for VO2max (p = 0.230), time to exhaustion (p = 0.143), HR at 9 km/h (p = 0.581), VO2 at 9 km/h, ventilatory threshold or RER; significant only for lactate at 9 km/h (p = 0.014), where both post-training values at two and three a week were lower than at one a week (p < 0.001).
Within the whole sample (main effect of time): VO2max up (F = 8.924, p = 0.005, η² = 0.190), time to exhaustion up (p < 0.001), HR at 9 km/h down (p < 0.001); the post hoc tests were run on the time effect, not per arm, so the paper does not show which arms changed on their own for VO2max.
Lactate within arms: lower after training in all three arms (one a week p = 0.011; two a week p < 0.001; the third p value is printed against "1/wk" again, a typo).
Ventilatory threshold (L/min) and RER: no main effects or interaction (VT 1.62-1.74 L/min, RER 0.84-0.85 throughout).
Individual response: SDs of change and individual data were not reported, so nothing can be said about responders at any frequency.
Withdrawals: 4 of 45, all "due to an inability to tolerate the training load" (one at two a week, three at three a week); the paper prints their age, height and body mass but no fitness data and makes no comparison with the completers. The "completion rate" is retention in the study; attendance at individual sessions was not reported.
Mechanisms
Measured: lactate after 8 min at 9 km/h fell more at two and three sessions a week than at one, while oxygen cost, ventilatory threshold and RER at that speed did not change; submaximal heart rate fell in all arms with no evidence of a difference between them (interaction p = 0.581).
Authors' interpretation: once the weekly volume of high-intensity stimulus passes a threshold, central and peripheral adaptations are largely in place and a third session gives diminishing returns; at one a week the gaps between sessions are too long for adaptation to accumulate. More frequent sessions may add peripheral adaptations (mitochondria, lactate transporters) that show as lower lactate at a submaximal speed.
Authors' interpretation: the longer time to exhaustion at one session a week, with VO2max barely changed, reflects tolerance of discomfort, pacing of effort and familiarity with maximal testing rather than a bigger engine.
Inference: the one-a-week arm did 6 sessions, the three-a-week arm 18; with no interaction on VO2max at 41 men, the trial cannot separate "frequency" from "total number of sessions", and it was not built to.
Practical Application
Do two interval sessions a week beat one? In 41 active young men doing 4×4-min runs for 6 weeks, VO2max rose ~1% at one a week, ~9-11% at two or three; the between-arm test was not significant. B+, transfer.
A third weekly 4×4 added little over a second and cost retention (B+, means, not a significant contrast): VO2max +4.54 vs +5.19 mL/kg/min at two vs three a week, and 3 of 15 withdrew at three a week against 1 of 15 at two. This agrees with SYNTH-26's "do not add a third hard session to catch up" and with REF-1291 (recreational runners, the same session, also 2 ≈ 3). Measured in active students, not endurance athletes.
One session a week barely moved VO2max in 6 weeks (+0.68 mL/kg/min) but its mean incremental-test time rose (+32.5 s, our arithmetic; no per-arm test) and its lactate at an easy speed fell (p = 0.011). For a week where only one quality session fits, it holds something but is a thin stimulus (inference); in inactive adults one longer weekly session did raise VO2max (REF-1292, A, Tier 0).
Not from this paper; the base's own practice: the number of quality sessions an amateur endurance athlete should do follows SYNTH-26's decision table by sport, sessions a week and training status (usually 2 on four to seven sessions a week, 1 for a runner on three runs, 2-3 on eight or more; "Recovery decides the actual count"); this trial supports "2 rather than 3" in active young men over 6 weeks and nothing about longer blocks or trained athletes.
Limitations
The frequency effect on VO2max is not statistically shown. The group-by-time interaction was p = 0.230 (time to exhaustion p = 0.143); no per-arm test of VO2max is reported. The abstract's "VO2max increased markedly in both the 2- and 3-session groups" is a description of means.
Small arms, completers only. 12-15 per arm; the four who withdrew (all from the two- and three-a-week arms, for intolerance of the load) were dropped, which flatters the higher-frequency arms if those who quit were the least adapted.
Registration after the fact; no blinding. Registered on 2 December 2025, after the trial; outcome assessors knew the allocation.
Who. Men only, about 20 years old, from one sport university, active in team sports and the gym; the paper does not say whether their usual sport continued during the trial. The three-a-week arm started with the lowest VO2max (47.90 vs 49.04-49.78), which may have helped its gain.
Short and lab-only. Six weeks; VO2max and an incremental-test time on a treadmill; no race or time-trial outcome, no perceived fatigue, soreness, sleep or stress recorded (the authors' limitation).
Reporting. The Discussion gives "approximately 5-6 ml·kg-1·min-1" gains at two and three a week; the two-a-week gain in the paper's own means is 4.54 (our arithmetic). "Completion rate" means staying in the study, not attending sessions. A post hoc p value is printed under the wrong group label.
Funding and interests: no funding received; no conflicts declared.
How this entry was checked and changed
Why this level
parallel RCT, n = 41 analysed of 45 randomised, read from the full text (open access). Not A- despite 45 randomised: three arms of 12-15 make each pairwise comparison a small trial, the frequency question rests on a non-significant group-by-time test for VO2max and time to exhaustion, the analysis dropped the four who withdrew, assessors were not blinded and the trial was registered after it ended
Full text (checked 2026-10-27)
45 male university students; randomly assigned to 4 × 4 min HIIT once, twice, or three sessions per week (n = 15 per group) for 6 weeks (full text)
VO2max: 49.78 ± 4.03 to 50.46 ± 4.24 (1 session), 49.04 ± 4.71 to 53.58 ± 5.58 (2 sessions), 47.90 ± 4.40 to 53.09 ± 6.97 ml·kg-1·min-1 (3 sessions); group × time interaction F(2, 38) = 1.527, p = 0.230, η² = 0.074 (full text)
Time to exhaustion: no significant group × time interaction (F(2, 38) = 2.051, p = 0.143, η² = 0.097) (full text)
Blood lactate at 9 km/h: 2.23 ± 0.27 to 2.04 ± 0.18 (1/week), 2.26 ± 0.21 to 1.77 ± 0.14 (2/week), 2.11 ± 0.17 to 1.71 ± 0.12 mmol·L-1 (3/week); group × time interaction F(2, 38) = 4.745, p = 0.014 (full text)
Training completion rates were 100%, 93%, and 80% in the 1−, 2−, and 3−session groups; withdrawals were due to an inability to tolerate the training load; analysis on completers only (N = 41) (full text)
Retrospectively registered at the Chinese Clinical Trial Registry on December 2, 2025, ChiCTR2500113733; outcome assessors were not blinded (full text)
Funding: financial support was not received for this work (full text)