Few ingredients generate as many confident headlines in both directions as collagen. In recent months alone, a well-run knee osteoarthritis trial found no benefit for undenatured type II collagen, although other trials and reviews continue to report positive improvements for skin, joints and muscle outcomes. While it may seem that the science is conflicted, more often the disagreement comes down to how the various studies were designed, analyzed and pooled, not whether collagen supplementation has a significant effect. The studies below show where those differences come from, and what stronger evidence looks like. The CSA advocates interpreting the evidence with nuance, while continuing to raise the standards, which is how the category earns trust. 

Placebo-Controlled Randomized Clinical Study Using a Collagen Hydrolysate-Enriched Diet in Dogs with Osteoarthritis 

In this randomized, double-blind trial from Utrecht University, a collagen-enriched diet performed no better than placebo in dogs with osteoarthritis. Forty-five dogs with mild, X-ray-confirmed arthritis ate either a diet with hydrolyzed porcine collagen (about 500 mg per kg of body weight per day) or a placebo diet matched for calories and protein for 12 weeks. Owners rated mobility and pain using validated questionnaires, and researchers measured weight-bearing on a force plate. Of the 36 dogs that finished, both groups improved, with no significant difference between them on any measure. More dogs on collagen reached a meaningful gain in weight-bearing (47% vs. 32%), but that gap wasn’t statistically significant. 

The authors’ discussion contains much of the nuance. In veterinary arthritis studies, up to 40-70% of animals in placebo groups have been reported to improve, in line with what the authors saw here. Owners may enroll their dogs when symptoms flare, so some improvement follows naturally, and the extra attention that comes with being in a study can itself help. The authors are also open about the study’s limits, 8 of the 9 dropouts came from the collagen group, and more placebo dogs started with moderate rather than mild lameness, leaving them more room to improve. Additionally, the authors note that in human studies, collagen hydrolysate’s effects on pain and cartilage have taken more than 6 months to appear. 

Change in Knee Osteoarthritis Cartilage Detected by Delayed Gadolinium Enhanced Magnetic Resonance Imaging Following Treatment with Collagen Hydrolysate: A Pilot Randomized Controlled Trial 

A small human trial from Tufts Medical Center shows what a longer timeline and a more objective measure can add. Thirty adults aged 49 and older with mild knee osteoarthritis took either 10 g a day of collagen hydrolysate (Fortigel) or an identical placebo for 48 weeks. Rather than using pain scores, the main outcome was a specialized MRI technique (dGEMRIC) that estimates proteoglycan content, a marker of cartilage health. In two of the six cartilage regions measured, both on the shinbone (tibia), scores rose in the collagen group and fell in the placebo group. The gap persisted at 48 weeks but fell just short of significance, and the other four regions showed no consistent pattern. As in the dog trial, pain and function improved in both groups with no difference between them, which the authors put down partly to regression to the mean. The placebo group also started with worse symptoms, and participants were allowed to keep taking their usual pain medication, both of which could have masked a difference. The authors stress that as a 30-person pilot, the results aren’t definitive and call for a larger trial. 

So what is the takeaway? Both trials point to the same limitations around study duration, and the measurements used profoundly affect the results. In each, symptoms improved in the placebo group as well as the collagen group, so owner-reported and patient-reported outcomes couldn’t separate the two. A short, small symptom-based trial that comes up empty is inconclusive rather than proof that an ingredient doesn’t work. Equally, a 30-person pilot that detects a change in cartilage isn’t proof that it does. We support larger trials that run long enough for a slow-acting ingredient, with more objective measurements of the joint itself rather than relying on symptoms alone, and more deliberate separation of participants by arthritis severity. 

Key Takeaways: 

• In a 12-week trial of 45 dogs with mild arthritis, a collagen hydrolysate diet performed no better than placebo. Both groups improved, which the authors attribute largely to caregiver placebo effects, and the study’s small size and short timeline limited what it could detect. 

• In a 48-week pilot trial of 30 adults with mild knee osteoarthritis, 10 g a day of collagen hydrolysate improved an MRI marker of cartilage health in two of six knee regions at 24 weeks, while pain and function improved equally in both groups. 

• Short, symptom-based trials are vulnerable to placebo effects. Longer trials with objective measures of the joint are better placed to show whether collagen makes a difference. 

Blees, N. R., Teunissen, M., Smit, I. H., Verseijden, F., Meij, B. P., Prevo-Keijbets, E., van Vuure, C., Corbee, R. J., & Tryfonidou, M. A. (2026). Placebo-controlled randomized clinical study using a collagen hydrolysate-enriched diet in dogs with osteoarthritis. The Veterinary Journal, 320, 106946. https://doi.org/10.1016/j.tvjl.2026.106946 

McAlindon, T. E., Nuite, M., Krishnan, N., Ruthazer, R., Price, L. L., Burstein, D., Griffith, J., & Flechsenhar, K. (2011). Change in knee osteoarthritis cartilage detected by delayed gadolinium enhanced magnetic resonance imaging following treatment with collagen hydrolysate: A pilot randomized controlled trial. Osteoarthritis and Cartilage, 19(4), 399-405. https://doi.org/10.1016/j.joca.2011.01.001 

Bioactive Peptide Profiling in Collagen Hydrolysates: Comparative Analysis Using Targeted and Untargeted Liquid Chromatography–Tandem Mass Spectrometry Quantification 

When trials and reviews talk about “collagen,” they often treat it as a single ingredient. This lab study tested that assumption by analyzing 10 commercial collagen hydrolysate products. Using a newly developed mass spectrometry method, the researcher measured three of the most studied bioactive collagen peptides (Pro-Hyp, Hyp-Gly and Gly-Pro-Hyp), along with each product’s peptide size distribution and amino acid profile. 

On paper, the products looked similar, all containing amino acid profiles commonly seen in collagen. However, more precise measurements revealed significant differences in peptide length, which varied widely across the collagen products. These differences largely stem from hydrolysis conditions during manufacturing, which have additional implications for peptide bioactivity. For example, Hyp-Gly ranged from 10 to 84 mg/kg, an eightfold difference, and one product contained far more Gly-Pro-Hyp than the other products.  

The nuance is that two products can both be labeled “collagen peptides,” share the same amino acid makeup, and still deliver very different effects. A positive trial on one product doesn’t automatically carry over to another, and a review that pools many different products may blur real effects or hide them altogether. The author suggests that reporting peptide content could give consumers more detail, reinforcing the need for verified sources, doses, and characterization across the category. It also illustrates the risk of borrowed science. When a product leans on clinical results from a different collagen ingredient, the claim may not hold up, and in a category built on consumer trust, that gap can damage credibility for everyone. 

Key Takeaways: 

• Ten commercial collagen hydrolysates had similar amino acid profiles but very different peptide sizes and bioactive peptide levels, with Hyp-Gly varying eightfold. 

• Products with more small peptides tended to have higher levels of Pro-Hyp, Hyp-Gly and Gly-Pro-Hyp. 

• Because “collagen” covers products this different, results from one product, or pooled across many, should be read with care. 

Oztug, M. (2024). Bioactive peptide profiling in collagen hydrolysates: Comparative analysis using targeted and untargeted liquid chromatography-tandem mass spectrometry quantification. Molecules, 29(11), 2592. https://doi.org/10.3390/molecules29112592 

Enzymatic Hydrolysis of a Collagen Hydrolysate Enhances Postprandial Absorption Rate—A Randomized Controlled Trial 

If products differ on the shelf, do those differences survive digestion? To answer this, we will examine two RCTs. In the first, from Aarhus University in Denmark, 10 healthy men took 35 g of a collagen hydrolysate that had been further broken down with an added enzyme step, 35 g of the same collagen without that step, or water, on three separate days. Blood samples taken over the next 4 hours showed that both collagen drinks raised amino acid levels, but the enzymatically hydrolyzed version delivered glycine, proline and hydroxyproline into the blood faster and in greater amounts. 

Absorption of Bioactive Peptides Following Collagen Hydrolysate Intake: A Randomized, Double-Blind Crossover Study in Healthy Individuals 

The second study, a pilot crossover trial in 6 healthy adults, compared 10 g doses of collagen hydrolysates from fish, pig and cow, including two bovine products with different average molecular weights (about 2,000 and 5,000 Da). Free hydroxyproline, a common marker of collagen absorption, rose by a similar amount whichever product people took, and molecular weight made no difference. But when the researchers looked at specific collagen peptides, including Pro-Hyp, Hyp-Gly and Gly-Pro-Hyp, blood levels differed significantly depending on the animal source, and the authors also point to differences between individuals. 

Together, these studies show why measuring absorption can hide meaningful differences depending on the approach. By the most common marker, free hydroxyproline, the products in the second study looked interchangeable, but this is not the case when looking at individual peptides. Both studies were small and single-dose, and neither measured health outcomes, so they tell us about absorption, not benefit. Still, they reinforce that source and processing influence a product’s efficacy, and trials should account for that. 

Key Takeaways: 

• In 10 healthy men, an extra enzymatic hydrolysis step made glycine, proline and hydroxyproline from 35 g of collagen reach the blood faster and in greater amounts. 

• In a pilot study of 6 adults, fish, porcine and bovine collagen gave similar free hydroxyproline uptake but significantly different levels of specific bioactive peptides, depending on source. 

• Source and processing affect what reaches the bloodstream, which is one more reason results from different collagen products shouldn’t be assumed to be interchangeable. 

Skov, K., Oxfeldt, M., Thøgersen, R., Hansen, M., & Bertram, H. C. (2019). Enzymatic hydrolysis of a collagen hydrolysate enhances postprandial absorption rate—A randomized controlled trial. Nutrients, 11(5), 1064. https://doi.org/10.3390/nu11051064 

Virgilio, N., Schön, C., Mödinger, Y., van der Steen, B., Vleminckx, S., van Holthoon, F. L., Kleinnijenhuis, A. J., Silva, C. I. F., & Prawitt, J. (2024). Absorption of bioactive peptides following collagen hydrolysate intake: A randomized, double-blind crossover study in healthy individuals. Frontiers in Nutrition, 11, 1416643. https://doi.org/10.3389/fnut.2024.1416643 

Interindividual Variability in Food-Derived Peptide Bioavailability and Function: Peptide Metabotypes and Responder Phenotypes 

Even with a well-characterized product and a well-run trial, bio-individuality can be a major confounder that most collagen studies often fail to discuss. This review, published in Food & Function, draws on human digestion studies, peptide absorption research, microbiome work and clinical trials of several food-derived peptides, including collagen peptides. The authors argue that the same product doesn’t produce the same exposure in everyone. Gut physiology, intestinal enzymes, how peptides cross the gut wall, the microbiome, age, metabolic health, medications and everyday diet can all change which peptides end up in the blood, and in what amounts. 

To make that measurable, they propose two ideas. A “peptide metabotype” is a person’s consistent pattern of peptides in the blood after a standard dose, and a “responder phenotype” links that pattern to whether, and how much, the person benefits. They also outline what a trial built around this would include: a well-characterized product, an appropriate comparator, repeated measurements, a targeted panel of absorbed peptides and a physiological endpoint chosen in advance. 

The industry Takeaway is that most collagen trials report an average effect across everyone in the study. If some people absorb and respond to collagen peptides well and others don’t, a modest average can hide a meaningful benefit for a subset of people, and two trials in different populations can reach different conclusions about the same product. This review doesn’t settle that question for collagen, but it gives the field a way to study it.  

Key Takeaways: 

• Gut physiology, the microbiome, age, metabolic health, medications and diet can all change how much of a food-derived peptide, including collagen peptides, reaches the blood. 

• The authors propose measuring each person’s “peptide metabotype” and linking it to whether they respond, which could help explain why average trial results vary. 

• Trials that characterize the product, measure absorption and pre-specify outcomes would make collagen evidence far easier to interpret. 

Suttikhana, I., & Ashaolu, T. J. (2026). Interindividual variability in food-derived peptide bioavailability and function: Peptide metabotypes and responder phenotypes. Food & Function. Advance online publication. https://doi.org/10.1039/d6fo03963d