There are six randomised trials of collagen peptides against exercise-induced muscle damage. They divide almost perfectly in two, and the line that divides them is not the dose, the population or the exercise protocol.
Counting the literature, then reading it
A PubMed search for collagen crossed with muscle soreness, muscle damage or DOMS, filtered to the randomised-controlled-trial publication type, returns 12 records. Six of those administered collagen to exercising humans. The others are two vitamin D trials, a transdermal-estrogen study in postmenopausal women, a plasma-hydroxyproline study of eccentric cycling intensities, an aerobic-exercise trial in knee osteoarthritis, and a study of ceftiofur sodium biobullet administration on tenderness and tissue damage in beef round muscle, which matches because cattle have collagen and slaughtered muscle can be damaged.
This is the fifth time a raw count on this site has turned out to be roughly half unrelated records. Reading the six that remain is where the finding is.
The three that worked
| Trial | Design | Intake | Result |
|---|---|---|---|
| 2019, Newcastle | 24 recreationally active males, double-blind, 150 drop jumps | 20 g/day, 7 days before + 2 days after | Soreness not significant (p = 0.071), large effect size at 48 h; jump-height recovery p = 0.050 |
| 2021, Oklahoma | 15 resistance-trained males (7 vs 8), 5 × 20 drop jumps | 15 g/day, 7 days | Jump height maintained at 24 h in one arm, significant decline in the other; neither collagen biomarker moved |
| 2023, Japan | 20 middle-aged males, randomised crossover, 5 × 40 squats | 10 g/day, 33 days per period | Reduced soreness; registered UMIN000041441 |
Two things about that table deserve stating plainly rather than being left in the citations.
The headline trial did not reach significance on its primary soreness outcome. The 2019 Newcastle paper is the one cited everywhere for collagen and recovery, and its soreness comparison came out at p = 0.071 with jump-height recovery landing exactly on p = 0.050. What it reported was a large effect size at 48 hours in 24 people. That is a legitimate reason to run a larger trial. It is not a demonstration.
Two of the three carry gelatin-industry authorship. The 2019 trial lists two authors at Rousselot BVBA in Ghent, a gelatin and collagen-peptide manufacturer. The 2023 crossover was authored at Nippi Inc, with author affiliations naming the company's Gelatin Division and its Research Institute of Biomatrix. Both papers disclose this; the disclosure is the source of the observation, not a discovery about them. The 2021 pilot, which is university-only, is also the smallest — seven people in the treatment arm — and its own abstract labels the two groups inconsistently when describing which one declined.
The two that compared collagen with actual protein
The other two trials asked a harder question. Instead of collagen against a placebo, they ran collagen against the protein people already use.
KU Leuven, 2024. Young fit males, three weeks of unilateral eccentric knee-extensor training, double-blind and parallel-group, with total protein held constant at 45 g a day: one arm took whey alone, the other took 25 g of whey plus 20 g of collagen peptides. Forty-eight hours after the first session, maximal voluntary isometric and dynamic contraction were impaired by about 10% and creatine kinase had doubled — in both groups. Training improved countermovement jump by 8% and dynamic contraction by 10%, and raised P1NP by 10% — in both groups. No outcome differed between them. Substituting collagen for part of a whey dose improved neither muscle damage nor function.
Massey University, 2024. Thirty-three males took 25 g of collagen hydrolysate, 25 g of dairy protein, or an isoenergetic placebo immediately after 30 minutes of downhill running at a minus-15% gradient, then daily for three days. Soreness, countermovement jump, isometric mid-thigh pull, maximal voluntary contraction, running economy, creatine kinase, interleukin-6 and high-sensitivity C-reactive protein all changed significantly over time — the protocol worked — and no group or interaction effect reached significance for any of them. The authors declare no conflicts of interest and record that the funders had no role in design, analysis or the decision to publish.
What the split actually says
Set the six trials side by side and the pattern is uncomfortable to state but easy to check:
- The trials that found a benefit compared collagen with nothing, and two of the three were co-authored by companies that sell gelatin.
- The trials that compared collagen with another protein found no benefit, and both declare no manufacturer involvement.
A placebo-controlled trial of a protein supplement answers a question nobody is really asking. The choice facing someone using collagen for recovery is not collagen versus no protein; it is collagen versus the whey or dairy protein they would otherwise have taken. That comparison has been run twice, with different designs and different damage protocols, and it came back null both times.
None of this touches the connective-tissue evidence, which is a separate literature with better results — the gelatin and vitamin C work and the joint trials are covered on our pages on peptides for recovery and peptides for joints and tendons. What collagen does to a tendon and what it does to a sore quadriceps are different claims, tested by different experiments, and only one of them keeps failing. The related question of whether collagen adds muscle rather than repairing it is handled separately, on collagen peptides and muscle growth, and the amino-acid arithmetic behind both is on collagen versus whey protein.
This is research journalism. Amounts are reported with their sources. Nothing here is a recommendation.
Sources
- Clifford T, Ventress M, Allerton DM, et al. The effects of collagen peptides on muscle damage, inflammation and bone turnover following exercise. Amino Acids 2019 — PMID 30783776
- Dietary collagen peptides alleviate exercise-induced muscle soreness in healthy middle-aged males: a randomized double-blinded crossover clinical trial. 2023 — PMID 37133292 · registry UMIN000041441
- Effects of Collagen Peptides on Recovery Following Eccentric Exercise in Resistance-Trained Males — A Pilot Study. Int J Sport Nutr Exerc Metab 2021 — PMID 33186897
- Partly Substituting Whey for Collagen Peptide Supplementation Improves Neither Indices of Muscle Damage Nor Recovery of Functional Capacity During Eccentric Exercise Training. Int J Sport Nutr Exerc Metab 2024 — PMID 37922892
- Comparing the Effects of Collagen Hydrolysate and Dairy Protein on Recovery from Eccentric Exercise: A Double Blind, Placebo-Controlled Study. Nutrients 2024;16(24):4389 — PMID 39771010 · PMC11678417