"Protein for recovery" sounds like one claim, but the trials split it into three things that do not move together: how fast strength comes back, how sore the muscle feels, and what the blood markers of damage show. Protein does something for the first, arguably something for the third, and nothing measurable for the second — and the most recent large analysis found the positive results cluster in trials that gave the protein group extra calories as well. This page reports what the trials measured; it is research journalism, not nutritional advice, per our editorial standards.
Three endpoints, three answers
The largest meta-analysis focused on exactly this question pooled 29 studies of protein taken around muscle-damaging resistance exercise (Pearson, Hind & Macnaughton 2023, PMID 36513777):
| What was measured | Time after exercise | Pooled effect (Hedges' g, 95% CI) | Reading |
|---|---|---|---|
| Isokinetic maximal strength | 24 h | 0.639 (0.116 to 1.162) | Favours protein |
| Isokinetic maximal strength | 48 h | 0.447 (0.104 to 0.790) | Favours protein |
| Isokinetic maximal strength | 72 h | 0.569 (0.136 to 1.002) | Favours protein |
| Isometric maximal strength | 96 h | 0.563 (0.232 to 0.894) | Favours protein |
| Creatine kinase | 48 h | 0.836 (−0.001 to 1.673) | Interval touches zero |
| Creatine kinase | 72 h | 1.335 (0.294 to 2.376) | Favours protein |
| Muscle soreness | all | no effect | No difference |
Two details in that table are easy to miss. First, the creatine kinase result the authors describe as large at 48 hours has an interval running from −0.001 — it is borderline, not established, until 72 hours. Second, soreness, the thing most people mean by "recovery", did not move at all. Creatine kinase, the blood marker, is a separate thing from creatine the supplement, a confusion covered on creatine in bloodwork.
The same team's own trial found nothing
The Durham group behind that meta-analysis then ran a randomised trial of its own. Twenty-four untrained men and women did one session of leg resistance exercise and took either milk protein or an energy-matched control, 17 doses over four days. Muscle damage was mild and was not attenuated by milk protein; strength, leg circumference, range of motion, soreness, pressure-pain threshold and creatine kinase were all tracked for a week (Pearson, Macnaughton & Hind 2023, PMID 36827652).
A null result in 24 people does not overturn a pooled effect, but it matters which kind of control it used. That is the thread the next analysis pulls.
The calorie problem
A protein shake compared with water, or with a zero-calorie placebo, adds two things: protein and energy. Zhao and colleagues' 2026 Bayesian meta-analysis — 75 randomised trials, 1,206 athletes, registered on PROSPERO — found protein supplements beat placebo on endurance and strength and preserved glycogen resynthesis better than carbohydrate. Then it split the trials by whether energy intake was matched between groups, and reported that all of the statistically significant effects came from trials in which it was not (Zhao et al. 2026, PMID 41433039).
The authors' conclusion is unusually direct for a meta-analysis: the observed benefits "may not be attributable to protein per se." Put plainly, part of what looks like a protein effect may be a more food effect — which is still real for someone under-eating, but is not something protein does that other calories cannot.
The army review saw the same split a decade earlier
A 2014 systematic review from the US Army Research Institute of Environmental Medicine read 27 articles and found "no apparent relationship" between recovery of muscle function, soreness ratings and blood markers of damage when protein was taken around a single bout of exercise, and insufficient evidence that post-exercise protein lowered soreness or damage markers at all (Pasiakos, Lieberman & McLellan 2014, PMID 24435468).
It named two situations where effects became more evident: protein taken after daily training sessions over time, rather than around one bout, and participants in negative energy or nitrogen balance — soldiers in the field, athletes in a deficit. The second point is the same finding Zhao's analysis reached with different data twelve years later: protein's recovery benefit is clearest where food is short.
What this means for the claims on the tub
- "Faster recovery" is supported in the narrow sense of maximal strength returning a little sooner in the days after unaccustomed damaging exercise, in pooled trials.
- "Less soreness" is not supported by the pooled evidence.
- "Because of the protein" is weakened by the calorie-matching finding: the controlled comparisons are where the effect thins out.
- Source differences — whey against plant, collagen against whey — are a separate question. The collagen-and-whey comparisons are covered on collagen versus whey protein and the collagen recovery trials, which split by who funded them, on collagen peptides for muscle recovery. The comparison with creatine as a supplement choice is on creatine versus protein powder, and the peptide side of the recovery question on peptides for recovery.
What is not established
- Whether protein matters once total calories are equal. The 2026 analysis suggests the effect shrinks or vanishes; it did not find a significant energy-matched effect.
- The best timing. "Peri-exercise" in the pooled trials covers before, during and after; the trials do not isolate a window.
- Effects in trained lifters doing familiar sessions. Most damage studies use unaccustomed eccentric or resistance exercise, often in untrained people, to create measurable damage.
- Anything about soreness beyond the null. No pooled analysis found here reports protein reducing it.
Sources and dates
- Pasiakos SM, Lieberman HR, McLellan TM. Effects of protein supplements on muscle damage, soreness and recovery of muscle function and physical performance: a systematic review. Sports Med 2014;44:655–70 — PMID 24435468
- Pearson AG, Hind K, Macnaughton LS. The impact of dietary protein supplementation on recovery from resistance exercise-induced muscle damage: a systematic review with meta-analysis. Eur J Clin Nutr 2023;77:767–83 — PMID 36513777
- Pearson AG, Macnaughton LS, Hind K. Milk protein ingestion does not enhance recovery from muscle-damaging resistance exercise in untrained males and females: a randomized controlled trial. Appl Physiol Nutr Metab 2023;48:455–68 — PMID 36827652
- Zhao S, Zhang X, Liang T, et al. The effectiveness of protein supplements on athletic performance and post-exercise recovery — a Bayesian multilevel meta-analysis of randomized controlled trials. J Int Soc Sports Nutr 2026;23:2605338 — PMID 41433039
- All abstracts read 2026-09-23.