Peptides

Collagen peptides: what the randomised trials actually show for skin, joints, hair and nails

By the US Health Digest editorial team · August 6, 2026

Collagen is the one part of the peptide market with a large randomised human evidence base — and it gives a different answer for each thing it is sold for. Skin: dozens of trials report gains in instrument-measured hydration, elasticity and wrinkle depth, but the most recent meta-analysis found them in industry-funded and lower-quality trials and absent in independently funded and higher-quality ones. Joints: a separate literature, graded very low quality, whose most-quoted pooled figure mixes hydrolysate with a 40 mg undenatured-type-II product. Nails: one 25-person open-label study with no placebo group. Hair: one randomised placebo-controlled trial of collagen alone, measuring shaft diameter, not hair counts or shedding. Collagen is not a weight-loss product.

What a collagen peptide actually is

A peptide is a short chain of amino acids — chemistry, not a category of medicine. As our pillar on what peptides are sets out, the word spans three markets with no shared evidence standard: prescription drugs such as the GLP-1 medicines, gray-market vials such as BPC-157, and the supplement aisle. Collagen is the aisle: bovine hide, fish skin, pig skin or chicken cartilage broken enzymatically into short fragments and sold as hydrolysed collagen, collagen hydrolysate or collagen peptides. Nearly every trial tests one named, proprietary hydrolysate.

The digestion objection, taken seriously

Eaten collagen becoming skin collagen is not how digestion works. A 2016 paper in Journal of Agricultural and Food Chemistry reports that collagen degrades randomly in the gut and "is not readily cleaved into bioactive peptides". In a 2025 randomised crossover study, free hydroxyproline, glycine and proline peaked 60 to 120 minutes after a single dose in water, with the fragments Pro-Hyp, Hyp-Gly and Gly-Pro-Hyp measurable in plasma; five of its eight authors work for Rousselot, a collagen maker. So amino acids and small fragments do reach the blood — but a fragment in plasma is pharmacokinetics, not a clinical effect.

How this evidence base is built

• Small and short. A 2026 review found 25 randomised trials with sample sizes from 13 to 236; most in a 2023 tabulation ran 2 to 12 weeks.
• Proprietary. A positive result belongs to the branded hydrolysate tested, not the category on the shelf, and does not transfer to an unspecified powder.
• Surrogate outcomes. Skin trials measure corneometry, cutometry and profilometry. A significant change in an instrument reading is a change in an instrument reading, not how anyone looks.
• Risk of bias. The 2026 review found "the majority of included trials were rated as having a high risk of bias"; the 2023 meta-analysis "identified several biases" in its own included trials.

Skin: the largest evidence base, and the fault line through it

The pooled analyses are favourable: a 2023 Nutrients meta-analysis of 26 trials in 1,721 participants found significant gains in hydration and elasticity (both p < 0.00001), and a 2021 meta-analysis of 19 trials in 1,125 participants, 95% women, agreed.

Then The American Journal of Medicine asked what the others had not. In September 2025, Seung-Kwon Myung and Youngjoo Park pooled 23 randomised trials in 1,474 participants, searched to June 14, 2024. Across all 23, collagen significantly improved all three outcomes. Split by funding source, trials not funded by pharmaceutical companies showed no effect on any of the three, while funded trials did. Split by quality, high-quality studies showed no significant effect in any category; low-quality ones improved elasticity only. Their conclusion: "There is currently no clinical evidence to support the use of collagen supplements to prevent or treat skin aging." Subgroup analyses can mislead, but this one is invisible in every pooled total. The abstract gives subgroup direction and significance, not pooled effect sizes with confidence intervals, and the full text is paywalled, so we print none.

In 2014, 114 women aged 45 to 65 took 2.5 g daily of a named hydrolysate (VERISOL) or placebo for 8 weeks and instrument-measured eye wrinkle volume fell 20% relative to placebo; two of six authors are affiliated with the Collagen Research Institute. In 2024, 85 women took 5 g of collagen peptides or maltodextrin for 84 days, with gains in dermis density and skin moisture; four of its five authors are Rousselot employees. Disclosed industry authorship is not misconduct — it is the pattern Myung and Park isolated.

Joints: one literature, several different products

The strongest independent synthesis is a 2018 British Journal of Sports Medicine meta-analysis by Xiaoqian Liu and colleagues covering 20 supplements across 69 randomised trials. Collagen hydrolysate was one of seven with large (effect size > 0.80) effects on pain at short term — but of that overall short-term result the authors wrote that "the quality of evidence was very low," and that "no supplements were identified with clinically important effects on pain reduction at long term."

The number quoted most often comes from a 2019 International Orthopaedics meta-analysis: collagen cut the total WOMAC index (weighted mean difference −8.00; 95% CI −13.04 to −2.95) and the VAS score (−16.57; 95% CI −26.24 to −6.89), while stiffness improved (−0.41; 95% CI −0.74 to −0.08) but WOMAC pain (−0.22; 95% CI −1.58 to 1.13) and function (−0.62; 95% CI −5.77 to 4.52) did not.

Then read what is in that pool. A 2020 scoping review tabulating it lists five included trials: Benito-Ruiz 2009, Kumar 2015, Lugo 2016, McAlindon 2011 and Schauss 2012 — two of collagen hydrolysate (10 g a day in one), one of collagen peptides from pork skin and bovine bone, one of a 2 g hydrolysed chicken cartilage extract, and the 40 mg undenatured type II trial below. That average spans products that are not interchangeable. It is not an estimate for hydrolysed collagen powder.

Undenatured type II collagen is not a collagen peptide

The most common error in consumer coverage. Undenatured type II collagen (UC-II) comes from chicken sternum cartilage, is deliberately not hydrolysed, and is dosed at 40 mg a day — about a sixtieth of a 2.5 g hydrolysate serving. In a 2016 Nutrition Journal trial, 191 volunteers took UC-II 40 mg, glucosamine plus chondroitin, or placebo for 180 days; UC-II reduced total WOMAC versus placebo (p = 0.002) and versus glucosamine/chondroitin (p = 0.04). Two of three authors were employed by InterHealth Nutraceuticals, which supplied UC-II. A tub of hydrolysed powder is not what that trial tested — and that trial is one of the five in the 2019 pool.

Hair and nails: what exists, and what does not

Nails. The study cited almost everywhere is a 2017 paper in Journal of Cosmetic Dermatology: 25 participants took 2.5 g of a named peptide (VERISOL) daily for 24 weeks, with a reported 12% increase in nail growth rate and 42% fewer broken nails. Its own methods call it "open-label, single-center" — no placebo, no randomisation, two of six authors from the Collagen Research Institute — so it cannot separate the supplement from ordinary variation or expectation. A 2025 review lists collagen among nail-supplement ingredients "shown to improve the clinical appearance of nails" while noting "manufacturers are not required to carry out efficacy and safety studies before marketing."

Hair. Searching PubMed via the NCBI E-utilities API on August 6, 2026, we located one randomised placebo-controlled trial of collagen on its own with a hair outcome: in 2026, 114 women aged 20 to 50 with thigh cellulite and self-reported hair thinning took 1,000 mg a day of a low-molecular-weight fish collagen peptide or placebo for 24 weeks, and hair shaft diameter rose significantly at week 24. Cellulite was the headline outcome, diameter is neither hair count nor shedding, and two of five authors work for the ingredient's maker. On density and shedding we found nothing randomised testing collagen alone. The near misses bundle it with other ingredients: a 2023 assessor-blinded trial of a tablet with 300 mg fish collagen plus taurine, cysteine, methionine, iron and selenium; a 2024 placebo-controlled trial of hydrolysed collagen with vitamin C, run by the maker's own staff, in which the 27.6% rise in hairs counted was not statistically significant; and an uncontrolled 2025 study of 38 women given personalised kits including marine collagen.

Use case by use case

What US regulation does and does not guarantee

FDA's consumer Q&A states that "under DSHEA, FDA does not have the authority to approve dietary supplements before they are marketed," and "FDA does not test dietary supplements before they are sold to consumers." Its power is post-market: it "has the authority to take action against any adulterated or misbranded dietary supplement product after it reaches the market."

Structure/function claims — "supports healthy skin" — are not pre-approved by FDA; the manufacturer holds its own substantiation, notifies FDA within 30 days, and prints a disclaimer that FDA has not evaluated the claim and the product is not intended to diagnose, treat, cure or prevent any disease. Advertising is the FTC's: its Health Products Compliance Guidance (December 2022) requires "competent and reliable scientific evidence" that is "sufficient in quality and quantity… when considered in light of the entire body of relevant and reliable scientific evidence" — the clause the funding-source finding puts in play.

Doses, sources and what is in the tub

In the 2023 Nutrients review's included trials, daily doses ranged from 372 mg to 12 g, clustering at 1, 2.5, 5 and 10 g over 2 to 12 weeks. Fish was the source in 14 of the 26 trials; nine specify no source at all. Hydration effects varied by source, but for elasticity there were no significant differences between sources (p = 0.21). Marketing marine collagen as superior runs ahead of that.

Contamination is documented for protein powders generally, with no collagen-specific dataset. A 2020 Toxicology Reports assessment by three Cardno ChemRisk authors cites a Consumer Reports analysis in which three daily servings exceeded heavy-metal limits proposed by US Pharmacopeia, and a follow-up finding elevated heavy metals in 40% of 133 products; its own EPA-based modelling found a hazard index below 1.

What to ask before buying

• Which named hydrolysate was tested, and is it in this tub?
• Hydrolysate at grams per day, or UC-II at 40 mg? If the joint claim rests on the UC-II trial, the product must be UC-II.
• Who funded the study cited at me? In the 2025 meta-analysis, funding source flipped the answer. The label's own disclaimer, required by law, says FDA has not evaluated the claim.
• Anything else going on? If you are pregnant, managing a condition, or on prescription medicines — including the prescription peptides we cost separately — ask your clinician.

Questions readers ask

Does eating collagen turn into collagen in my skin?

No. Published research states that collagen degrades in the gastrointestinal tract and is not readily cleaved into bioactive peptides; what reaches the bloodstream is free amino acids plus small di- and tripeptides, and whether those do anything to skin is a separate question.

Is collagen for joints the same product as collagen for skin?

Often not. Hydrolysed collagen is taken at grams per day. Undenatured type II collagen is a different, deliberately unhydrolysed molecule taken at 40 mg a day, and is what the osteoarthritis trial most often cited for joints tested.

Does collagen help hair growth?

Searching PubMed on August 6, 2026, we found one randomised placebo-controlled trial of collagen on its own with a hair outcome: 1,000 mg a day for 24 weeks increased hair shaft diameter in 114 women whose main complaint was cellulite. On hair density or shedding we found nothing randomised testing collagen alone; those studies bundle collagen with other ingredients, so no effect can be attributed to it.