REF-1329 · Tripp et al., 2025 · Scandinavian Journal of Medicine & Science in Sports · Paper · PubMed · Free full text
Is training twice a week as good as four times for the same weekly time?
For VO2max in 28 untrained adults over 8 weeks, yes: gains differed by 0.7 mL/kg/min (CI −1.5 to 2.9, margin 3.5). The riding was heavy plus intervals, not easy.
Main caveat: Who: sedentary-to-recreationally-active adults, VO2max about 43 mL/kg/min; trained riders, masters and anyone above 45 were not studied.
Tripp and colleagues (MacInnis lab, University of Calgary, with Gibala) asked whether training on two days a week is worse than on four when the week's time and intensity are the same.
Twenty-eight sedentary-to-recreationally-active adults (14 men, 14 women, 18-45 y, VO2max about 43 mL/kg/min) were randomised, stratified by sex and VO2max, to 8 weeks of supervised ergometer cycling: four sessions a week (two continuous, two HIIT) or two sessions on back-to-back days in which each session merged two training units (one continuous day, one HIIT day), so each session was twice as long.
Continuous riding was at 50% of the gap between GET and RCP power (25 → 35 min in HF, 50 → 70 min in WW, plus warm-up and cool-down); HIIT was 4-min intervals at 110% of RCP power with 3-min recoveries at 50 W (3-4 intervals in HF, 6-8 in WW). Weekly totals were identical by design: 132 min in weeks 1-2 and 156 min from week 3, with 99.2% of sessions completed.
VO2max rose by about 10% in both arms; the 95% CI of the difference in improvement was −1.5 to 2.9 mL/kg/min (p = 0.43), inside the pre-registered 3.5 mL/kg/min (1 MET) margin, so twice a week was not inferior.
GET and RCP (as VO2 and power), peak power, haemoglobin mass and blood volumes, NIRS muscle oxidative capacity (rising in the first 4 weeks only), severe-domain time to task failure and the heart-rate, lactate and RPE responses to 30 min of heavy riding all improved with no frequency effect; the differences were that only the four-times-a-week arm reduced the fall in potentiated twitch force (peripheral fatigue) after heavy and severe exercise, and that body mass fell slightly in the twice-a-week arm only (interaction p = 0.049; no time point differed on post hoc testing).
Our reading: in untrained adults, the weekly dose, not how it is sliced, drove 8-week fitness gains; the trial says nothing direct about easy riding or trained athletes.
Protocol
Design: randomised (concealed allotment in blocks of 4, stratified by sex and baseline VO2max), two parallel arms, non-inferiority test on the change in relative VO2max; margin 3.5 mL/kg/min (1 MET) set before the study; sample size 26 for α 0.05, power 0.9 and an SD of change of 3 mL/kg/min. Ethics: University of Calgary CHREB (23-0467).
Participants: 30 recruited; 2 (1 man, 1 woman) dropped out after 2-3 weeks for scheduling reasons and were excluded; n = 28 analysed (14 per arm, 14 men and 14 women). Healthy, 18-45 y, not endurance trained (no more than 4 sessions or 3 h a week). Baseline body mass 70.5 (HF) and 72.3 kg (WW); body fat 23.6% and 27.1%.
Training (Supporting Information Table S1): HF 4 sessions a week on any days; WW 2 sessions on back-to-back days. Continuous: Δ50% GET-RCP power, 25, 30, 35 min in weeks 1-3 for HF (CONT 50, 60, 70 min for WW), plus 5 min warm-up and 2 min cool-down. HIIT: 4 min at 110% of RCP power, 3 min at 50 W between, 3 then 4 intervals for HF (6 then 8 for WW), after an 8-min warm-up.Intensity raised about 5% in week 4, reset to the new thresholds after mid-point testing and raised again in weeks 6-7 by tolerance. Weekly total 132 min (weeks 1-2), 156 min (weeks 3 and 5-8), 120 min in week 4 (one HIIT unit replaced by a test). Longest continuous ride: about 35 min in HF, about 70 min in WW; the authors give "~75 min" as the WW session length.
Setting: supervised on Tacx NEO smart trainers (two on a Velotron), Polar H10 strap; four participants did up to 8 of 32 units remotely. Habitual activity was to be kept; some did other activities 1-3 days a week (climbing, coaching, weightlifting).
Tests (0, 4 and 8 weeks): step-ramp-step incremental test (VO2max, peak power, GET and RCP corrected for the VO2 mean response time and slow component); NIRS oxidative capacity of vastus lateralis (0, 4, 8 weeks); DXA (0, 8); haemoglobin mass by CO rebreathing, in duplicate (0, 8); a performance trial of 30 min at Δ70% GET-RCP (143 ± 33 W at baseline) then a time-to-task-failure bout at 85% of peak power (224 ± 43 W), with knee-extensor neuromuscular function (MVC, potentiated twitch, 10:100 Hz doublet ratio, voluntary activation), gas exchange, lactate, RPE and fatigue.After training the trial was repeated at the same absolute and the same relative power.
Analysis: non-inferiority by the 95% CI of the difference in VO2max change; two-way repeated-measures ANOVA or linear mixed models (group × training) for secondary outcomes, Sidak post hoc; exploratory sex × training analysis; α 0.05.
Funding and interests: internal funding from the University of Calgary Faculty of Kinesiology and an NSERC Discovery Grant to the senior author; the first author held NSERC and University of Calgary scholarships. The authors declare no conflicts of interest.
Key Results
Means ± SD before → after 8 weeks; Tables 1-4, abstract and Supporting Information:
| Outcome | 4 sessions/week (HF) | 2 sessions/week (WW) | Group × training |
|---|---|---|---|
| VO2max, mL/kg/min (Table 1 labels; see the note below) | 43.5 ± 6.4 → 47.8 ± 6.4 | 42.3 ± 6.2 → 47.3 ± 6.7 | p = 0.772 |
| VO2max, L/min | 3.07 ± 0.72 → 3.37 ± 0.72 | 3.04 ± 0.59 → 3.35 ± 0.67 | p = 0.921 |
| Peak power, W | 257 ± 47 → 288 ± 52 | 266 ± 53 → 295 ± 53 | p = 0.159 |
| Haemoglobin mass, g (change) | 754 ± 185 → 765 ± 202 (11 ± 42) | 771 ± 203 → 790 ± 189 (18 ± 27) | p = 0.596 |
| NIRS oxidative capacity k, s−1 (abstract) | 0.036 ± 0.011 → 0.041 ± 0.010 | 0.034 ± 0.008 → 0.045 ± 0.015 | p > 0.05 |
| Severe time to task failure, same power, min (abstract) | 3.5 ± 0.9 → 7.7 ± 2.8 | 3.7 ± 1.6 → 8.6 ± 3.2 | p > 0.05 |
| 30 min heavy, same power: heart rate, bpm | 170 ± 14 → 157 ± 15 | 166 ± 13 → 153 ± 13 | p = 0.839 |
| 30 min heavy, same power: blood lactate, mM | 6.0 ± 1.5 → 3.6 ± 1.5 | 5.7 ± 2.2 → 3.3 ± 1.3 | p = 0.972 |
| 30 min heavy, same power: energy from fat, % | 31.5 ± 9.9 → 37.8 ± 9.6 | 30.8 ± 12.4 → 36.4 ± 14.9 | p = 0.913 |
| Potentiated twitch fall, pre-exercise → after TTF, % | −55.6 ± 8.0 → −40.2 ± 9.0 | −48.9 ± 14.9 → −46.4 ± 14.1 | p = 0.003 |
Primary outcome: difference in VO2max improvement 95% CI −1.5 to 2.9 mL/kg/min, not significant (p = 0.43), upper bound below the +3.5 margin → twice a week was not inferior. The Discussion gives the gap between arms as 0.7 mL/kg/min and the gain in both as about 10%.
A number that does not add up: the abstract gives WW 43.5 ± 6.5 → 47.8 ± 6.4 and HF 42.3 ± 6.2 → 47.3 ± 6.7; Table 1 gives the same rows to the opposite arms (HF 43.5 → 47.8, WW 42.3 → 47.3).Table 1's own absolute VO2max and body mass reproduce its labels (3.07 L/min ÷ 70.5 kg = 43.5; 3.37 ÷ 70.5 = 47.8; our arithmetic), while the Methods' CI direction (HF minus WW, upper bound against the margin) fits the abstract's; the Introduction instead says the lower bound must not reach the margin, which would fit Table 1.Which arm gained the 0.7 mL/kg/min more cannot be settled from the paper; non-inferiority holds either way, since both bounds lie inside ±3.5. The NIRS and time-to-task-failure rows above come from the abstract and carry its arm labels, which could not be checked against a table.
Thresholds: VO2 at RCP rose 46.9 → 52.9 and at GET 34.3 → 39.3 mL/kg/min (both arms pooled, no interaction; Table S2 prints mL/kg/min, but these must be per kg fat-free mass, since per kg body mass RCP would exceed VO2max of about 43 — our reading); threshold power also rose in both.
Blood volumes (pooled): plasma volume +~5%, red-cell volume and Hb mass +~2%.
Muscle oxidative capacity: up after 4 weeks (pooled 0.035 → 0.040 s−1), not further at 8 weeks (0.043); the authors give ~23% after 8 weeks.
Same relative power after training: time to task failure did not change (pooled 3.6 → 4.0 min, p = 0.927), so riders held their duration at the higher new power.
Sex: no sex × training interaction for VO2max, thresholds, oxidative capacity, blood volumes or time to task failure (exploratory, not powered).
Individual response: change scores are shown only in figures, and with no untrained control arm the spread cannot be split into true response and measurement noise (REF-1274).
Mechanisms
Measured: both arms raised oxygen delivery (Hb mass, blood and plasma volume) and muscle oxidative capacity (NIRS), lowered heart rate, ventilation, carbohydrate oxidation and lactate at a fixed heavy power, and resisted overall knee-extensor fatigue (MVC) better; only the four-times-a-week arm reduced the potentiated-twitch fall (p = 0.010 after the heavy bout, 0.003 after task failure).
Authors' interpretation: for a given weekly volume, frequency is not an important determinant of short-term gains; the size of the exercise stress is set mostly by intensity, and acute muscle fatigue, metabolites and kinase signalling rise non-linearly with session duration, so a session twice as long is not twice the stimulus — yet the WW arm still matched HF with half the exposures.More frequent exposure to the peripheral fatigue of interval sessions may explain HF's twitch-force resilience. They speculate that the WW arm, used to longer rides, might show more durability in a longer task (not tested; the performance trial lasted 30 min plus the severe bout).
Inference: the frequency effect in REF-465 (6 > 4 > 2 sessions a week) comes from trials in which more sessions usually also meant more weekly time; here, with time matched, frequency added nothing measurable for VO2max, thresholds or blood volume.
Inference: the result concerns heavy-domain and interval work in untrained adults. For easy riding, where the stimulus per minute is smaller, whether 2 × 70 min equals 4 × 35 min was not tested.
Practical Application
Is training twice a week as good as four times for the same weekly time? For VO2max in 28 untrained adults over 8 weeks, yes: gains differed by 0.7 mL/kg/min (CI −1.5 to 2.9, margin 3.5). The riding was heavy plus intervals, not easy.
For an untrained or returning rider, fewer and longer sessions are a legitimate shape for the week: at the same weekly time and intensity (here about 2.2-2.6 h), two sessions on back-to-back days gave 8-week VO2max gains not inferior to four, and no evidence of a difference in thresholds, blood volume or muscle oxidative capacity (secondary outcomes, not powered; B+, transfer from untrained adults).The authors' caution: a beginner may not tolerate sessions of about 75 min on back-to-back days at first, so build session length gradually.
Peripheral fatigue resistance was the main thing frequency bought: only four sessions a week reduced the potentiated-twitch fall after hard riding (secondary outcome, small trial). Not from this paper; the base's rule (SYNTH-84 rule 2, on REF-1097): stacking costs quality, so when volume is spread or doubled, keep the key session first or on its own day.
Easy riding is not covered: the continuous sessions sat halfway between the two thresholds. Not from this paper; the base's own practice: for the easy part of the week, choose the arrangement by what the athlete will repeat and by the event (the long session is optional for events up to ~2 h; for 2-5 h events spreading is fine in base, with a regular long ride in the last 8-10 weeks, SYNTH-84).
Limitations
Who: sedentary-to-recreationally-active adults, VO2max about 43 mL/kg/min; trained riders, masters and anyone above 45 were not studied.
What: heavy-domain continuous riding (Δ50% GET-RCP) plus intervals at 110% RCP power; no easy (below-GET) riding, and no arm compared 2 vs 4 sessions of easy work.
How long and how much: 8 weeks at 132-156 min a week; whether low frequency stays equal over months, or at 6-10 h a week, is unknown.
No control arm: an active comparator only, so the authors cannot fully attribute every change to the training, and individual responses cannot be judged.
Power: sized for VO2max; secondary null interactions may be type II errors (the authors give partial eta squared to show possible undetected effects).
Not a strict two-day week: some participants did other activities 1-3 days a week; four did part of the training remotely; training was supervised, which the authors say lowers ecological validity and removes adherence as an outcome.
Sex: not powered for sex differences; menstrual cycle and contraceptive use not recorded.
Reporting: the abstract and Table 1 assign the two VO2max rows to opposite arms (and give 6.5 vs 6.4 for one SD); Hb mass agrees with Table 2; the abstract's per-arm NIRS and time-to-task-failure numbers have no per-arm table but average to the pooled Table S2 means (our arithmetic).
Funding and interests: University of Calgary Faculty of Kinesiology internal funding and an NSERC Discovery Grant; no conflicts declared. The trainers (Tacx, a Garmin brand) and heart-rate straps (Polar) were equipment, not named as funders.
How this entry was checked and changed
Why this level
parallel RCT, n = 28 (14 per arm), read from the full text and Supporting Information (open access, CC BY-NC-ND 4.0); the rubric's B+ for a parallel RCT of 20-39. Not higher: an active comparator without a no-training control, 8 weeks, powered for VO2max only (the authors warn the secondary comparisons may be underpowered), and an untrained sample
Full text (checked 2026-10-28)
Twenty-eight sedentary-to-recreationally-active adults aged 18–45 years (14 males and 14 females), 8 weeks on a cycle ergometer, four or two sessions weekly, intensity and volume matched (full text)
The bounds of the 95% confidence interval (−1.5 to 2.9 mL/kg/min) did not include the a priori non-inferiority margin of +3.5 mL/kg/min; p = 0.43 (full text)
Continuous sessions at 50% of the difference between the GET and RCP power outputs; HIIT 4 min at 110% of RCP power output with 3 min of recovery at 50 W (full text)
Weekly plan: 132 min Total in weeks 1-2, 156 min Total in weeks 3 and 5-8 and 120 min Total in week 4, with WW continuous sessions of 50, 60 and 70 min, Supporting Information Table S1 (full text)
Table 1 VO2max (mL/kg BM/min): HF 43.5 ± 6.4 to 47.8 ± 6.4, WW 42.3 ± 6.2 to 47.3 ± 6.7; the abstract gives WW: 43.5 ± 6.5 vs. 47.8 ± 6.4 and HF: 42.3 ± 6.2 vs. 47.3 ± 6.7 (full text)
The protocols are much closer than the margin (0.7 mL/kg/min); both improved VO2max by ~10% (full text)
Pt force, pre-exercise to post-TTF: HF −55.6 ± 8.0 to −40.2 ± 9.0%, WW −48.9 ± 14.9 to −46.4 ± 14.1%, interaction p 0.003 (full text)
Average adherence of 99.2% session completion (full text)
Individuals beginning with fewer, longer sessions may not tolerate exercise sessions of ~75 min on back-to-back days (full text)
Funded by internal funding from the University of Calgary Faculty of Kinesiology and an NSERC Discovery Grant (RGPIN-2018-06424); the authors declare no conflicts of interest (full text)