113 clinical trials, 8,000 people, one answer: does collagen actually work?

For years, every collagen conversation on the internet has ended the same way. Someone says their skin and joints feel better. Someone else replies that collagen is "just expensive protein" that your stomach tears apart. Both sides shrug and go home. As a company that sells the stuff, we've been more honest than most about that argument, and we've been open that we'd love better data.

In June 2026, better data arrived, and it's not a small study we cherry-picked. Researchers at Anglia Ruskin University pulled together every systematic review they could find on collagen supplementation: 16 reviews covering 113 randomised controlled trials and roughly 8,000 participants, published in Aesthetic Surgery Journal Open Forum and covered everywhere from ScienceDaily to The Conversation and TIME. An umbrella review like this is about as high up the evidence pyramid as nutrition science gets.

Here's what it found, including the parts that don't flatter us.

What the evidence supports

Skin. Across the pooled trials, collagen supplementation improved skin elasticity and hydration compared with placebo. This backs up what smaller individual trials had been showing for a decade; one frequently cited RCT by Proksch and colleagues found significantly better skin elasticity after 8 weeks of just 2.5 to 5g of collagen peptides daily (DOI). The umbrella review's larger data pool means this is no longer resting on a handful of industry-friendly studies.

Joints. The review found reduced pain and stiffness in people with osteoarthritis. For the many of you who arrive at this blog from searches about creaky knees rather than crow's feet, this may be the more useful half of the verdict.

Time. One of the most practical findings: benefits grew with longer supplementation. The people who took collagen for longer saw more, which matches exactly what we tell customers in our week-by-week results timeline: judge it at 12 weeks, not 12 days.

What the evidence doesn't support

The same review found little effect on post-exercise muscle soreness. So if you were buying collagen purely as a recovery aid for DOMS, the data says save your money or buy it for other reasons (for muscle recovery specifically, creatine has far better evidence). Collagen also isn't a complete protein, so it shouldn't be counted toward your protein target the way whey or chicken is.

We think telling you this plainly matters. A supplement company that claims its product does everything is telling you it hasn't read the research. Collagen has two well-supported jobs, skin and joints, and a long tail of maybes.

How to buy collagen that matches the research

If the umbrella review has persuaded you to try it, or to restart the tub you abandoned, three things separate the products that resemble the trials from the ones that just borrow the word "collagen".

First, hydrolysed peptides. The trials used hydrolysed collagen, meaning the protein is pre-broken into small absorbable peptides. Every Nutrabytes collagen is fully hydrolysed; we've written a plain-English explainer on what that means if you like the mechanics.

Second, a real dose. Positive skin trials used roughly 2.5 to 10g daily and joint trials clustered around 10g. Plenty of high-street capsules contain a few hundred milligrams. Our liquid marine collagen and bovine collagen powder both deliver 10,000mg per serving, the top of the studied range, so the dose question is simply off the table.

Third, a format you'll still be using in week twelve. Since the review found benefits build over months, the best collagen is whichever one you'll take daily without willpower. The liquid is a pre-measured daily shot with vitamin C built in (which contributes to normal collagen formation in its own right). The unflavoured powder disappears into coffee or a smoothie. If you're torn between fish and beef sourcing, our marine vs bovine comparison settles it honestly in five minutes.

The bottom line

The largest evidence review ever conducted on collagen landed on the side of the people who kept taking it: real, measurable improvements in skin elasticity, hydration and joint comfort, growing over time, at doses of a few grams a day. Not a miracle, not snake oil. A supplement that does two specific things, backed by 113 trials, as long as you take enough of it for long enough.

That last clause is the whole game, and it's the one thing no study can do for you.

Quick answers

What did the 2026 collagen study actually find? The Anglia Ruskin umbrella review (16 systematic reviews, 113 RCTs, ~8,000 people) found improved skin elasticity and hydration and reduced osteoarthritis pain and stiffness, with benefits increasing over longer use, and little effect on post-exercise soreness.

How much collagen should I take? Successful trials mostly used 2.5 to 10g daily. All Nutrabytes collagens provide 10g per serving.

How long until collagen works? Weeks, not days. Most positive trials ran 8 to 12 weeks; the review found longer use produced larger benefits. Full timeline here.

Is marine or bovine collagen better? Both were represented in the successful trials. Marine edges it for skin-focused goals, bovine for value and joints. Full comparison.

Does collagen help sore muscles after the gym? The review found little effect on post-exercise soreness. For recovery, creatine and adequate complete protein have better evidence.

Food supplements should not replace a varied, balanced diet. If you are pregnant, breastfeeding or taking medication, check with your GP before starting any new supplement.