Cold Plunges: What the 2023-2025 RCTs Show

The ice bath has become a symbol of discipline and biohacking: athletes, entrepreneurs and bloggers show it off as a universal upgrade for the body. The trouble is that the promise is one thing — muscle, recovery, immunity, mental clarity — while controlled studies measure different effects separately. And those effects sometimes contradict each other. Let's sort it out, point by point, with what the randomized trials and meta-analyses of 2023–2025 actually showed.

Does cold really speed up recovery?

Here the data are on cold's side. A network meta-analysis of 55 randomized controlled trials, published in Frontiers in Physiology in 2025 (Wang and colleagues), compared different "doses" of immersion by temperature and duration. Of the 55 studies, 42 measured delayed-onset muscle soreness (DOMS), 36 measured jump readiness, and 30 measured creatine kinase, a biochemical marker of muscle damage.

The best performer was a moderate protocol: 10–15 minutes at 11–15 °C. It reduced muscle soreness with an effect of SMD −1.45 (95% confidence interval from −2.13 to −0.77) and significantly lowered creatine kinase. In other words, after a hard bout cold really does blunt soreness and speed up the feeling of readiness. The authors explicitly recommend moderate temperatures and durations rather than an extreme icy shock.

So why does cold harm muscle growth?

This is where the central paradox lies. The very inflammation and stress that cold suppresses are the signal that triggers muscle growth after a strength session. Suppress the signal, and the body adapts worse.

This was confirmed by Piñero's meta-analysis with the telling title "Throwing cold water on muscle growth" (European Journal of Sport Science, 2024), which pooled 8 studies. Strength training on its own produced a gain in lean mass with an effect of SMD 0.36 (credible interval 0.10–0.61) — reliable, if small. The cold-water-immersion-plus-lifting pairing produced only 0.14 (interval from −0.08 to 0.36), i.e. trivial to small. The direct comparison favored training without cold. Studies with muscle biopsies showed blunted hypertrophy of type II fibers — exactly the ones responsible for strength and size.

An important caveat: the authors rated the quality of the included studies as "fair-to-low," and the cold protocols were typical for ice baths — around 10 °C for 10–20 minutes. But the direction of the effect is consistent: if your goal is mass, don't get cold right after lifting.

And what about immunity and mood?

This is exactly where the hype is greatest and the evidence smallest. A systematic review and meta-analysis in PLOS One (2025, Cain and colleagues) pooled 11 studies and 3,177 participants at water temperatures of 7–15 °C.

The picture turned out to be "by the clock." Right after immersion, inflammation didn't drop — it actually rose (SMD 1.03), and an hour later it was still elevated (SMD 1.26) — an acute response to cold stress. But after 12 hours, subjective stress fell (SMD −1.00). For immunity the meta-analysis found no acute changes; the most cited result was an observational 29% reduction in sick-day absences from work among cold-shower practitioners. For mood, no reliable effect emerged; modest gains showed up for sleep and quality of life.

The key word is "observational." The 29% drop in sick days was not obtained in a strict randomized trial, so it's a prompt for further study, not a proven cause. The authors note outright: the protocols are heterogeneous, women are underrepresented in the samples, and long-term follow-ups are almost nonexistent.

So, do it or not?

The answer depends on your goal, and that's the whole point. Cold is not a universal upgrade but a tool with a specific profile. It's good when you need to recover quickly between bouts and relieve soreness. It gets in the way when you're building mass and getting cold right after lifting. And when it comes to immunity and the mind, it still falls under "maybe, but unproven" — and it certainly doesn't replace sleep, nutrition and regular training.