Does collagen actually work? Who paid for the studies

Five meta-analyses of collagen and skin have been published since 2019, and most report benefits for hydration, elasticity and wrinkles. Only one of them, published in the American Journal of Medicine in 2025 across 23 randomized trials and 1,474 participants, analyzed the results by funding source.

In that stratification, the non-industry-funded studies showed no effect on hydration, elasticity or wrinkles. Restricting to high-quality studies showed no effect in any category. The authors' conclusion is unambiguous: there is currently no clinical evidence to support the use of collagen supplements to prevent or treat skin aging.

This is not a claim that researchers cheated. It is the ordinary, well-documented tendency of industry-funded trials to produce industry-friendly results, showing up in a literature where nearly every trial has a manufacturer somewhere in it.

Enough that a reader checking conflict-of-interest statements would be misled, because in trial after trial the conflict statement and the funding statement contradict each other. The pattern is consistent enough to list.

The flagship muscle trial in sarcopenic men declares "there are no conflicts of interest" and acknowledges that part of the costs were paid by Gelita AG. The postmenopausal bone-density trial declares no conflict of interest and acknowledges that part of the study was financially supported by Gelita AG. The premenopausal muscle trial declares no conflict of interest and was in part funded by Gelita GmbH. The 2023 muscle-remodeling trial declares no conflict of interest and was funded by the Collagen Research Institute in Kiel. The Achilles tendinopathy pilot declares no conflicts of interest while its PubMed grant field names Gelita A.G.

Elsewhere the relationship is in the author list rather than the funding line. Two of six authors on the most-cited skin trial work at the Collagen Research Institute, and the product tested is Gelita's. A co-author on another skin trial is an employee of the company that makes the tested ingredient. Every author on a 2024 osteoarthritis trial works for the gelatin company whose own brand was tested. One trial's conflict statement discloses that the funders "were allowed to make suggestions in phrasing."

Each of these may satisfy its journal's rules. The aggregate effect is that this literature looks more independent than it is.

This is where collagen has its most credible signal, and it still needs a ruler held against it.

A 2023 meta-analysis of knee osteoarthritis found a pooled pain effect of -0.58 standard mean difference (95% CI -0.98 to -0.18). It rests on four trials and 507 patients, and its authors state that all of them were considered to have a high risk of bias. A 2018 review in the British Journal of Sports Medicine placed collagen hydrolysate among supplements with large short-term effects on pain, then added that no supplement showed clinically important effects at long term and that the quality of evidence was very low.

Now the ruler. The minimum clinically important difference on an 11-point knee pain scale is 1.8 points, which is 18 mm on a 100 mm visual analogue scale. The two best-powered collagen knee trials produced between-group differences of 5.6 mm and 6.3 mm. Both were statistically significant. Both are roughly a third of the smallest change a person would notice.

One more detail worth carrying: in that meta-analysis, adverse events came out at an odds ratio of 1.66 (95% CI 0.99 to 2.78), rated very low certainty. It does not establish harm, and it is not nothing.

The most useful analysis here is a 2024 systematic review in Sports Medicine covering 19 studies and 768 participants, funded by the University of Vienna with no external funding and no commercial relationships. It gives collagen its fairest hearing, and the results split cleanly.

Fat-free mass with resistance training came out at a standardized mean difference of 0.48 (95% CI 0.22 to 0.74), rated GRADE moderate. That is the strongest finding in the collagen literature and it deserves to be stated plainly. Two caveats travel with it: six of the eight studies measured body composition by bioimpedance, which struggles to resolve differences of about a kilogram and is sensitive to hydration; and independent isotope studies suggest the effect is not specific to collagen.

Tendon told a different story. Tendon morphology improved (0.67) at very low certainty, with publication bias detected. Tendon mechanical properties did not move at all: 0.05, not significant, confidence interval -0.38 to 0.48. Maximal strength was 0.19 at low certainty. The honest sentence is that tendons may get bigger without getting stiffer or stronger.

The mechanism evidence is where this gets difficult for the category. When researchers used stable isotopes to measure protein synthesis directly, collagen did not perform. Two 15 g doses a day for a week did not raise myofibrillar or muscle connective protein synthesis rates. In older women, whey raised muscle protein synthesis and collagen did not. In older men, whey and pea both raised integrated synthesis rates while it remained unchanged with collagen.

There is a straightforward reason. Collagen is an incomplete protein: gelatin contains no tryptophan at all and only about 2.5 percent leucine, the amino acid that triggers muscle protein synthesis. Vital Proteins says so on its own site.

The headline trial randomized 131 postmenopausal women, 102 of whom completed 12 months on 5 g a day. Spine T-score changed by +0.1 versus -0.03 for placebo (p=0.030) and femoral neck by +0.09 versus -0.01 (p=0.003). A T-score change of 0.1 is roughly one percent of bone mineral density, which sits inside the repeat precision of a DXA scanner. No collagen trial has ever measured a fracture.

The four-year follow-up sometimes cited alongside it retained 31 of the original 131 participants, was open-label, and had no control group. It cannot demonstrate efficacy.

Half of this is well established. Collagen peptides survive digestion: after a 9 to 23 g dose, hydroxyproline-containing dipeptides such as Pro-Hyp appear in human blood at 20 to 60 nmol/mL, peaking at one to two hours and halving by four. That is real, replicated, and measured by mass spectrometry. Worth noting for anyone paying a premium: plain gelatin raises blood Pro-Hyp about as well as hydrolysate does.

The other half is not established in humans. The evidence that these peptides reach skin or cartilage at a working concentration comes from rats, in a study whose sole author was affiliated with a supplement company. A 2017 paper whose title promises "and Skin" measured plasma in humans and skin in mice, with five of nine authors at a cosmetics company.

And the numbers do not line up. The laboratory experiment showing Pro-Hyp makes skin fibroblasts migrate used 200 nmol/mL. Peak human blood levels are 15.5 to 60.65 nmol/mL, for about two hours. The dish concentration is three to thirteen times higher than a human bloodstream ever reaches.

The field has also corrected itself in a way worth respecting. In 2020, researchers discovered that the fetal bovine serum used in these cell experiments already contained 70 to 100 micromolar hydroxyprolyl peptides, which had been masking results, and wrote that this raises doubts about conclusions from prior research.

No US authority has ever reviewed a bovine or marine collagen peptide supplement before sale. There is no GRAS notice for collagen peptides, hydrolyzed collagen, collagen hydrolysate, marine collagen or bovine collagen in the FDA's inventory, and because collagen was marketed before 1994 it never triggered a new-ingredient notification either.

Alpha-gal syndrome. Bovine collagen is a mammalian product, and bovine collagen alpha-1 (VI) chain has been identified as a galactose-alpha-1,3-galactose-bound allergen in beef. A 2026 CDC analysis found alpha-gal IgE seroprevalence of 31.2 percent among blood donors in Arkansas and 26.0 percent in Missouri, with up to 450,000 US people affected. Beef is not a major allergen under US labeling law, so bovine collagen carries no allergen declaration requirement. Fish is a declared allergen, and fish collagen anaphylaxis is documented: a confirmed case in a nine-year-old after two collagen gummies.

Oxalate, for anyone who forms kidney stones. Gelatin is about 9.17 g hydroxyproline per 100 g, so 10 g of collagen delivers roughly 0.92 g of hydroxyproline (our arithmetic). A 5 g gelatin load raised urinary oxalate in one study while 2 g did not, which puts typical marketed doses at or above the threshold that mattered. This is specific to collagen rather than a general protein effect: raising dietary protein from 0.6 to 1.8 g/kg/day did not increase urinary oxalate. One contradicting trial in five healthy men found no change in oxalate excretion after 30 g of gelatin. And a direct search finds no study has ever tested whether collagen supplements raise stone risk.

What is not a real worry. BSE. Gelatin was never considered a prohibited cattle material, the FDA regards the risk of human exposure through food as negligible, and a 2024 EFSA assessment put the probability of no new BSE case arising from oral exposure to ruminant-bone gelatine at 99 to 100 percent, explicitly extending that to collagen.

Heavy metals, reported carefully. A 2020 advocacy-group analysis of 30 collagen products found measurable lead in 37 percent and measurable arsenic in 64 percent. It was not peer reviewed, sampled Amazon best-sellers, and was disputed by two industry trade bodies, all of which are interested parties. The benchmark most often misquoted alongside it is California's Proposition 65 lead level of 0.5 mcg a day, which is a warning threshold, not a toxicity threshold. The FDA sets no binding maximum for lead in dietary supplements.

Two FDA warning letters in 2021 went to collagen sellers, both for disease claims rather than product safety. One cited claims that collagen has a therapeutic role in preventing and treating osteoporosis and osteoarthritis. The other cited claims that shark collagen components directly attack and kill cancer cells and shrink tumors. Both concluded that the claims made the products drugs under the Federal Food, Drug, and Cosmetic Act.

On the advertising side, the FTC sued the marketer of a collagen joint product in December 2019 over claims including "the only collagen clinically proven to reduce pain by 95%." It ended in an $821,000 judgment, with $774,755 returned to 13,221 consumers, an average of $58.60 each. That is one verified case, not a pattern.

Read from each brand's own page on 2026-08-22, with the monthly figure our arithmetic at the 10 g a day used in the joint trials: Sports Research $21.92, Vital Proteins $24.11 at the larger size, NOW Foods $34.27, Thorne Collagen Plus $51.54. That is a 2.4-fold spread for the same daily dose. Thorne's is a combination formula rather than plain collagen, so it is not comparable per gram.

Two things to check on any label. Scoop size is not collagen content: an 11 g scoop may deliver 10 g or 11 g of collagen depending on the brand. And third-party testing claims vary widely, with none of the products above carrying NSF Certified for Sport or Informed Sport.

If you are taking collagen for your skin, the best-designed analysis of that question found no effect once funding source and study quality were accounted for. If you are taking it for knee pain, the effect is real in pooled analysis and about a third of what a person would notice, from four trials all rated high risk of bias. If you are taking it alongside resistance training for body composition, that is the one moderate-certainty finding, and independent isotope work suggests any protein would do it better.

Collagen is a food, it is cheap at the low end, and there is no category-level harm signal. If you enjoy it and it fits your budget, that is a defensible reason to keep taking it. What the evidence does not currently support is paying a premium for it on the strength of the marketing.

Before starting collagen if you form kidney stones, have alpha-gal syndrome or a fish allergy, or are pregnant. And if you are taking it for joint pain that is limiting what you can do, that is worth a proper assessment: osteoarthritis has treatments with far better evidence than any supplement, starting with strength training and weight management.