Cold-Water Immersion Eases Post-Exercise Soreness, but Evidence Review Reveals Redundant Research and Uneven Benefits

Cold-water immersion, the icy bath that has become a near-obligatory ritual for everyone from professional soccer players to weekend runners, does appear to deliver on at least one of its promises: reducing the self-reported muscle soreness that peaks a day or two after hard exercise. But a sweeping new analysis of the review literature finds that beyond soreness, the evidence for cold-water immersion as a recovery tool is far shakier than its popularity suggests, and that much of the apparent scientific consensus rests on the same small pool of primary studies being counted over and over again.

The analysis, published as an overlap-aware umbrella review in Sports Medicine – Open, was led by Weibao Liang and colleagues and prospectively registered in PROSPERO. The team searched six electronic databases, including PubMed, Web of Science, SPORTDiscus, the Cochrane Library, Embase and Scopus, from their inception to 30 January 2026, looking for systematic reviews, meta-analyses and network meta-analyses of cold-water immersion applied after exercise. From 679 identified records, the researchers screened 334 and assessed 79 full-text reports before settling on 14 reviews: eleven systematic reviews with pairwise meta-analysis, two network meta-analyses and one systematic review without meta-analysis.

What sets this umbrella review apart is its treatment of duplicate research. Rather than simply pooling the pooled estimates from different reviews, the authors quantified how much the reviews reused the same underlying trials. They collapsed multiple reports, arms, subgroups and timepoints from the same publication into single study families, then calculated the corrected covered area, a metric in which values above 15 percent signal very high overlap. Across 437 study-review inclusion cells drawn from 165 unique primary-study families, the overall corrected covered area was 12.68 percent, classed as high. Sixty percent of the study families appeared in at least two reviews, and 62.24 percent of all inclusions were redundant. Because these overlapping estimates are statistically dependent, the authors declined to perform a second-order meta-analysis, which would have treated dependent numbers as independent and produced confidence intervals that were too narrow.

The degree of redundancy varied by outcome and was often extreme. The corrected covered area reached 15.03 percent for delayed onset muscle soreness, 23.08 percent for perceived recovery, 17.10 percent for jump performance, 19.64 percent for strength, 21.79 percent for power, 16.75 percent for creatine kinase, 25.00 percent for C-reactive protein and 15.38 percent for lactate. It was 14.29 percent for sprint performance, 6.25 percent for perceived fatigue and 0.00 percent for interleukin-6, though the researchers caution that the zero value for interleukin-6 reflected only two clinically heterogeneous reviews sharing no study families, not an invitation to pool. Heart rate variability could not be assessed because a single review covered that outcome, contributing only narrative evidence from 12 studies, of which seven reported statistically favorable findings compatible with acute parasympathetic reactivation.

Against this backdrop of redundancy, one signal stood out. Delayed onset muscle soreness showed the most consistent favorable pattern across the review set, with overall direction-harmonized standardized mean differences ranging from −1.04 to −0.37, all favoring cold-water immersion. Yet even this consistency comes with caveats. The four reviews contributing the overall estimates were not independent: one shared three to six soreness-study families with each of the other three. The largest estimate, drawn from a review of trained soccer players, was −1.04 with a 95 percent confidence interval of −1.91 to −0.17, but its 95 percent prediction interval stretched from −4.12 to 2.03, crossing zero, meaning that a future study in a similar setting could plausibly find no effect. Several time-specific estimates in other reviews also crossed zero, and one review reported no clear difference at 96 hours or beyond.

Objective performance measures told a much less flattering story. Sprint findings showed no consistent benefit, with one review reporting an immediate estimate that actually favored the comparator, and the soccer-specific review finding no clear overall 20-meter sprint effect. Jump, strength and power results were heterogeneous: some reviews reported early impairment in jump performance within the first hours after immersion, others found no clear differences, and the most favorable estimates came with prediction intervals compatible with null effects in future settings. Creatine kinase, an indirect marker of muscle membrane disruption with large inter-individual variability tied to training status, muscle mass, sex and ethnicity, varied by review and context, and the authors emphasize that a lower creatine kinase concentration does not necessarily mean better functional recovery. C-reactive protein showed no consistent benefit, interleukin-6 evidence was sparse, and lactate was null immediately after immersion in the reviews that measured it.

Methodological quality compounded the uncertainty. Using AMSTAR 2, a validated appraisal instrument for systematic reviews, the authors rated two of the 14 reviews as low confidence and 12 as critically low; none achieved moderate or high confidence. Common critical flaws included the absence of lists of excluded studies, lack of prospective protocol registration, inadequate integration of primary-study risk of bias into conclusions, and incomplete investigation of reporting bias. These weaknesses in the source reviews carry forward into any umbrella analysis built upon them, and the authors flag them explicitly as a limitation of their own synthesis.

The review also examined how comparator choice and protocol details shape the apparent effects. Passive recovery, active recovery, thermoneutral water immersion, contrast water therapy and body cryotherapy are not interchangeable benchmarks, and estimates shifted depending on which was used; in one review, cold-water immersion reduced soreness relative to active recovery at one hour but not at 24 or 48 hours. Several reviews reported favorable results in protocol strata around 5 to 15 degrees Celsius for 5 to 15 minutes, typically applied immediately after exercise, but the authors stress these are descriptive observations drawn from review-level subgroups, meta-regressions or dependent network nodes. Protocols below 5 degrees Celsius were rarely studied, and the findings should not be read as an optimized temperature-and-duration prescription. The physiological rationale for the soreness benefit remains plausible: cooling lowers tissue temperature and nerve conduction velocity, alters nociceptive input and raises pain threshold, while the hydrostatic pressure of immersion shifts blood centrally. But the authors note that soreness is a subjective outcome vulnerable to expectation effects, citing an experimental placebo-controlled study in which perceived recovery benefits after cold-water immersion were no greater than with a thermoneutral placebo.

The review’s conclusions are measured. Cold-water immersion was associated with the most consistent reduction in self-reported short-term post-exercise muscle soreness, but the clinical importance of that reduction is uncertain, and the apparent consistency partly reflects reuse of the same primary trials rather than independent replications. Evidence for objective performance, creatine kinase, lactate, C-reactive protein, interleukin-6 and heart rate variability was mixed, context-dependent or insufficient. The authors suggest cold-water immersion is best viewed as a targeted option for short-term soreness management rather than a uniformly effective recovery strategy, and they call for future reviews to preregister protocols, report complete study-to-outcome contribution tables, disclose funding, apply and integrate risk-of-bias assessments, and record participant training status, sex, comparator and protocol details precisely. They also note that women remain under-represented in this literature, that most reviews mixed training levels, and that only the soccer-focused review isolated a sport-specific population, whose findings should not be generalized to all athletes. For now, the ice bath keeps its place in the recovery toolbox, but chiefly as a pain-relief measure whose broader recovery claims the evidence does not uniformly support.

Subject of Research: Umbrella review of cold-water immersion for post-exercise recovery, quantifying primary-study overlap across systematic reviews

Article Title: Cold-Water Immersion for Post-exercise Recovery: An Overlap-Aware Umbrella Review

Article References: Liang, W., Ma, X., Sun, G., Liu, C., & Yan, X. (2026). Cold-Water Immersion for Post-exercise Recovery: An Overlap-Aware Umbrella Review. Sports Medicine – Open, 12(1), Article 159. https://doi.org/10.1186/s40798-026-01132-8

Image Credits: AI Generated

DOI: 10.1186/s40798-026-01132-8

Keywords: cold-water immersion, post-exercise recovery, delayed onset muscle soreness, umbrella review, primary-study overlap, corrected covered area, AMSTAR 2, sports medicine, meta-analysis, inflammation biomarkers, athletic performance, systematic review quality