Effects of cold-water immersion at different body regions on post-exercise muscle damage recovery: a systematic review and meta-analysis
Participants
30 randomized controlled trials comparing a single acute CWI with seated passive rest; total participant number and training status are not given in the abstract.
Protocol
PRISMA; PubMed, Embase, Web of Science and the Cochrane Library searched from inception to 20 October 2025. Quality by the PEDro scale, risk of bias by RoB 2 (overall quality rated high). Stata-MP 18.0. Subgroup analysis: whole-body vs partial immersion, at follow-ups from 0 to 72 h. PROSPERO CRD420251171826.
Key Results
| Outcome | CWI vs control (Hedges' g, 95% CI) | P |
|---|---|---|
| CK (creatine kinase) | −0.24 (−0.37 to −0.10) — reduced | < 0.01 |
| DOMS (soreness) | −0.40 (−0.64 to −0.16) — reduced | < 0.01 |
| MVIC (maximal strength) | 0.08 (−0.08 to 0.23) — no benefit | > 0.05 |
| CMJ (explosive power) | −0.02 (−0.17 to 0.13) — no benefit | > 0.05 |
| CMJ immediately (0 h) after partial CWI | −0.94 — inhibited | < 0.01 |
Whole-body vs partial immersion: no significant difference at any follow-up time point (0-72 h; between-group P > 0.05) — the authors call this equivalence the core finding
The biochemical (CK) and subjective (DOMS) benefits were concentrated at 24 h
Publication bias (Egger's test) was detected for CK (P < 0.001) and DOMS (P = 0.025); after trim-and-fill the CK effect shrank from g −0.236 to −0.086 (95% CI −0.210 to 0.037) and lost significance — the authors say the CK results warrant cautious interpretation; DOMS stayed significant (corrected 2026-10-12, full text)
Mechanism
Not addressed in the abstract. Inference, not tested: cooling limits blood flow and metabolic activity in damaged tissue (less inflammation and swelling), which fits the CK and soreness effects; acute cooling of the working muscles fits the immediate drop in jump power.
Practical Application
Meta-analysis of 30 RCTs: a single cold-water immersion after damaging exercise lowered creatine kinase (g −0.24; not significant after trim-and-fill correction for publication bias (corrected 2026-10-12, full text)) and soreness (g −0.40) with the benefit concentrated at 24 h, but did nothing for the recovery of maximal strength (g 0.08) or jump power (g −0.02), and partial (lower-limb) immersion inhibited jump power immediately (g −0.94) — CWI makes you feel and test less damaged the next day; it does not bring strength or power back faster.
Legs-only immersion is as effective as a full-body bath at every time point from 0 to 72 h — a bin or tub for the legs is enough.
Do not use CWI right before anything explosive (sprints, jumps, a hard start): the authors advise caution "during intervals involving continuous explosive power output". It is for accelerating next-day recovery, not for between-effort recovery on the same day.
Not from this paper: REF-011 (Roberts 2015) — regular post-strength CWI blunts strength and hypertrophy adaptation; keep CWI away from strength sessions whose adaptation you want.
Not from this paper: REF-019 (Chauvineau 2021) — evening whole-body CWI and early-night slow-wave sleep, provisional.
Not from this paper; the base's own practice: water around 10-15 °C for 10-15 min (the abstract does not report the pooled temperature or duration).
Limitations
Protocols (temperature, duration, timing) varied across the 30 trials; the abstract does not report the ranges or a dose-response analysis
Participant training status is not described in the abstract; elite-athlete data are likely thin
CMJ was the explosive-power measure; no sport-specific power or endurance performance outcome
Publication bias detected for CK and DOMS; after trim-and-fill the CK effect lost significance (g −0.086, 95% CI −0.210 to 0.037), DOMS held (corrected 2026-10-12, full text)
Single acute immersion only; says nothing about repeated CWI over a training block (for that, REF-011)
How this entry was checked and changed
Content correction 2026-09-27
DOMS "g −0.42; CI −0.64 to −0.18" -> g −0.40; CI −0.64 to −0.16.
"CMJ immediate g −0.34 (0-10 h)" and "CMJ 24 h: significant recovery benefit" -> the pooled CMJ effect was g −0.02 (no benefit); the immediate inhibition was g −0.94 (P < 0.01) after partial CWI; there was no CMJ benefit at 24 h — the 24-h concentration concerns CK and DOMS.
"Partial immersion equally effective for explosive power recovery" -> no CWI mode improved explosive power; partial immersion was equivalent to whole-body on all outcomes.
"10-15 °C × 10-15 min supported by the meta-analysis", "avoid within 10 h", "≥12 h before next workout", Limitations "5-15 °C, 5-20 min", mechanism paragraph -> deleted or relabelled (base practice / inference); not in the abstract. Tier "Elite 4-5 (review, predominantly this population)" -> mixed, Sport "cycling, strength, sprint" -> general: the abstract describes neither.
Cross-reference descriptions checked against the cited entries' current content; wrong years, wrong entries and claims the cited entry does not make were corrected.
Full text (checked 2026-10-12)
The 30 trials (published 2008 to 2025) pooled 527 participants (498 male, 32 female) aged 15.6 to 50.3 years; 18 trials involved professional athletes and 12 healthy non-athletes (full text).
Damage protocols were eccentric or eccentric-concentric resistance exercise in 12 trials, HIIT (sprints, LIST/Yo-Yo, simulated matches) in 15 and aerobic endurance exercise in 3 (full text).
CWI was applied within 1 hour post-exercise; protocols ranged from 5.0 °C to 20.0 °C and from 6 min to 25 min, with lower-body immersion defined as water at or below the iliac crest (full text, Table 2).
CK: lower-body CWI reduced CK only at 24 h (g −0.30, P = 0.02), whole-body CWI did not reach significance at 24 h (g −0.43, P = 0.07) (full text, Table 4).
DOMS: whole-body CWI reduced soreness at 0 h (g −0.88, P = 0.03) and lower-body CWI at 24 h (g −0.52, P = 0.04); DOMS heterogeneity was I2 = 76.02%, falling to 53.71% when the Elias 2012 study (g –4.62) was removed (full text, Table 5).
CMJ: whole-body CWI at 0 h was g −0.35 (P = 0.09) and at 24 h g 0.26 (P = 0.05), with no significant effect at 48 or 72 h (full text, Table 6).
Correction: the entry says the main results held in sensitivity analysis with publication bias only "detected for CK"; the full text reports Egger's test bias for CK (P < 0.001) and DOMS (P = 0.025), and after trim-and-fill the CK effect shifted from −0.236 to −0.086 (95% CI: −0.210 to 0.037) and lost significance, so the authors say the CK results "warrant cautious interpretation" (full text).
Adverse events are listed as NR (not reported) for every included trial (full text, Table 2).