“Collagen” might be the single most successful word in the supplement aisle. It is in coffee creamers, gummies, beauty waters and protein powders. And because collagen is a protein, the marketing prefers the more scientific-sounding term: collagen peptides — short fragments of hydrolyzed collagen designed to be easy to absorb. It is the same peptide-buzzword energy as the rest of the wellness world, with one difference. Unlike most trendy ingredients, oral collagen peptides have a genuinely large pile of randomized controlled trials behind them, and two meta-analyses that pooled them.
Does drinking collagen work?
Yes, modestly, and mainly for skin. Pooled across 19 randomized, double-blind, placebo-controlled trials in 1,125 people, and again across 26 trials in 1,721 people, hydrolyzed collagen significantly improved skin hydration and elasticity compared with placebo; wrinkle measures improved too, but less consistently.[5][6] The effective window was 2.5 to 10 grams a day for 8 to 12 weeks, the effect size was small-to-moderate rather than transformative, and a large share of the trials were paid for by companies that sell collagen.
That is the whole answer in three sentences. The rest of this page is the part that decides whether it is worth your money: which measurements actually moved, how big the change was, what the funding problem does to the numbers, and the claims — muscle, joints, hair — that the skin evidence does not transfer to.
The obvious objection — and why it is not the whole story
Skeptics raise a fair point: when you swallow collagen, your gut breaks it down into amino acids and small peptides, the same as any other protein. It does not get couriered straight to your face. So how could it possibly reach skin? The answer that survives scrutiny is that not all of it is broken down. When volunteers drank gelatin hydrolysates, researchers could measure collagen-specific peptides appearing in their blood: the hydroxyproline-containing dipeptide prolyl-hydroxyproline (Pro-Hyp) rose from a negligible baseline to a peak of 20 to 60 nanomoles per millilitre of plasma one to two hours after the drink, falling to half that by four hours.[8] Those fragments are the plausible messenger: the working theory is that they act as signals nudging skin fibroblasts to build more collagen and matrix. The mechanism is still being argued over, which is exactly why the human outcome data matters more than the theory.
What the individual trials show
This is collagen’s strong suit. A randomized, double-blind, placebo-controlled trial by Kim and colleagues gave 64 people either placebo or 1,000 mg/day of low-molecular-weight collagen peptide for 12 weeks; the collagen group showed significantly better skin hydration by week 6, plus improvements in wrinkling and elasticity by week 12, with no adverse effects.[1] An earlier double-blind trial in 69 women aged 35–55 tested 2.5 g and 5 g/day of specific collagen peptides and found significantly improved skin elasticity versus placebo after 8 weeks — with the benefit still measurable four weeks after they stopped taking it.[2]
The pattern keeps repeating in newer and larger studies. According to PubMed, a 100-person randomized, double-blind, placebo-controlled trial of low-molecular-weight collagen peptides reported significant improvements in wrinkles, elasticity and hydration at 12 weeks.[3] And a 2024 trial of 85 women, layered on top of a standardized skincare routine, found that 5 g/day of collagen peptides significantly improved dermis density and skin moisture compared with placebo over 84 days.[4] Across these studies the doses cluster around 1–5 g daily, the timelines around 8–12 weeks, and the safety record is clean.
What happens when you pool every trial
Single trials are easy to cherry-pick, so the more useful question is what the whole literature looks like at once. Three systematic reviews have tried. A 2019 dermatology review found 11 randomized placebo-controlled trials in 805 patients, with eight of them using collagen hydrolysate at 2.5 to 10 g a day for 8 to 24 weeks, and concluded that oral collagen increased skin elasticity, hydration and dermal collagen density with no reported adverse events.[7] A 2021 meta-analysis went further and pooled 19 randomized, double-blind, controlled trials in 1,125 participants aged 20 to 70, finding favorable effects on hydration, elasticity and wrinkles, with hydration and elasticity holding up in subgroup analysis; the authors concluded that roughly 90 days of hydrolyzed collagen reduces measurable signs of skin aging.[5]
The largest pooled analysis so far, published in 2023, covered 26 randomized controlled trials in 1,721 patients and reported significant improvements in skin hydration (Z = 4.94, p < 0.00001) and elasticity (Z = 4.49, p < 0.00001) versus placebo. Its subgroup analysis found that hydration results varied with the collagen source and the duration of supplementation, but that elasticity did not differ significantly between sources (p = 0.21). The same authors noted that they identified several biases in the included trials and called for larger, better studies.[6]
Two things are worth noticing in those pooled numbers. The statistics tell you the effect is unlikely to be chance; they do not tell you it is visible in a mirror. And the population studied is narrow: 95 percent of the participants in the 2021 meta-analysis were women, mostly middle-aged, mostly with dry or aging skin at baseline. Nobody has shown that a 28-year-old with normal hydration gets anything at all.
Dose and timeline: the two numbers that matter
If you strip the evidence down to practical instructions, there are only two numbers. Dose: 2.5 to 10 grams a day of hydrolyzed collagen peptides, with 2.5 to 5 grams covering most of the positive trials. Duration: 8 to 12 weeks of daily use before you decide anything, because hydration tends to move first and elasticity and wrinkle measures lag behind it. Consistency beats dose here. A 10-gram scoop taken twice a week is a worse bet than 2.5 grams taken every morning, and nothing in the literature suggests that going above 10 grams buys more skin benefit — it mostly buys calories and shelf space.
Timing is the one question the trials cannot answer. Some brands insist on an empty stomach, others on taking it with vitamin C. The blood-peptide work shows Pro-Hyp peaking one to two hours after ingestion regardless, and no trial has compared timing head to head. Take it whenever you will actually remember it.
Types I, II and III: what the number on the tub means
The human body makes at least 16 types of collagen, but three dominate: types I, II and III.[9] That matters, because the tub you buy is sold on exactly that number and most shoppers have no idea what it refers to.
Type I is the workhorse of skin, bone, tendon and ligament, and it is the one every skin trial above used, usually hydrolyzed and often with some type III alongside it, since the two sit together in the dermis. If a product is aimed at skin, hair or nails, it should be a type I and III hydrolysate. Type II is a different animal: it is the collagen of cartilage, and it has its own research literature in knee osteoarthritis, where the undenatured form is used at about 40 mg a day and is thought to work through oral tolerance, an immune mechanism, rather than by supplying building material.[10] Note the dose: 40 milligrams, not 10 grams. Undenatured type II and hydrolyzed type I are not interchangeable products, and buying a joint formula for your face — or the reverse — is a category error the labels do little to prevent. We go through the joint, bone and muscle claims separately in our full collagen guide.
Marine, bovine, porcine or chicken
Source is the other thing the packaging shouts about, and it is the thing the evidence cares least about. Marine collagen (fish skin and scales) is type I, often with a slightly lower molecular weight; bovine is type I and III; porcine is type I and III; chicken is mostly type II and therefore a joint product. In the 26-trial meta-analysis, collagen source made no significant difference to skin elasticity, though it did shift the hydration subgroup results — and those trials were never designed to compare sources head to head, so the comparison is indirect at best.[6]
Which leaves the non-scientific criteria to do the deciding: fish allergy rules out marine, religious dietary rules rule out porcine, and nothing on the market is vegan, because collagen only exists in animals. Products advertised as “vegan collagen” are collagen builders — vitamin C, amino acids, sometimes silica — not collagen. Because every real option is an animal by-product, third-party testing for heavy metals and purity is worth more than any label claim about hydrolysis technology.
Powder, capsules or food?
The format question is mostly arithmetic. The trial doses are in grams, and a capsule holds roughly 500 mg of powder, so matching a 5-gram trial dose means swallowing about ten capsules a day. That is why capsules are the most expensive way to get a dose that works, and why most people who stick with collagen end up using plain unflavoured powder, which dissolves into coffee or yoghurt without taste. Collagen gummies are the weakest format of all: a typical gummy carries well under a gram, and the sugar is not free.
Food is a legitimate route with one caveat: the dose is unknowable. Gelatin is cooked collagen, and slow-cooked cuts, chicken skin, skin-on fish and a long-simmered bone broth all deliver real collagen peptides — the original blood-peptide experiments used gelatin hydrolysates, not branded powders.[8] But the collagen content of home-made broth varies enormously with bones, time and water, and most measured samples land well under the 2.5 grams a day the trials used. The deeper point is that collagen is not an essential nutrient at all: your body assembles it from glycine, proline and vitamin C, which is why total protein intake does more work here than any single scoop. If you have never put a number on that, our protein calculator is a faster way to find a gap than any supplement label.
The catch: who paid for the research
Here is the grown-up caveat. A large share of collagen skin trials are funded or run by companies that make or sell collagen ingredients. That does not make the results fake — these are real randomized, placebo-controlled, double-blind studies, the gold standard of design. But funding source is reliably associated with the direction of the conclusion in nutrition research: in a classic analysis of 206 nutrition articles, industry-funded studies were several times more likely to reach conclusions favorable to the sponsor’s interests than independently funded ones.[12] Applied here, the honest read is that the direction of the collagen evidence is genuinely positive while the size of the benefit probably sits at the optimistic end of reality. The 2023 meta-analysis reaching the same positive conclusion while explicitly flagging biases in its included trials is about as good a summary of the state of play as exists.[6]
Who actually benefits — and who should not bother
The trials point at a fairly specific person: a woman in her forties or older with dryness and measurable loss of elasticity, already using sunscreen, who will take a powder daily for three months. That is who was studied, and that is who the effect sizes belong to. If your skin concern is pigmentation or sun damage rather than dryness and laxity, collagen is the wrong tool entirely — that is sunscreen and retinoid territory, and the cross-linking damage behind stiff, leathery skin is a different process altogether.
Three groups can reasonably skip it. Anyone taking it for muscle: collagen is protein by definition but nearly devoid of leucine, the amino acid that triggers muscle protein synthesis, and in a randomized crossover trial in trained adults an equal dose of lactalbumin raised myofibrillar protein synthesis about 13 percent more than collagen peptides did.[11] If muscle is the goal, total protein and resistance training are the lever. Anyone taking it for joints without checking the type: that literature runs on undenatured type II at 40 mg, not hydrolysate at 10 g.[10] And anyone on a tight budget: at typical prices a year of collagen costs more than a year of sunscreen plus a retinoid, which have better evidence for the same goal. Collagen is a reasonable extra, not a foundation. The same ranking logic applies across the shelf, which we worked through in our guide to the longevity supplements actually worth your money.
How to tell whether it is working
Skin changes slowly and memory is unreliable, so run it like a small experiment. Take a photo in the same light and the same expression before you start, pick one thing you actually care about — morning tightness, how makeup sits, the crease depth you notice in that light — and do not re-evaluate for 12 weeks. Change nothing else in your routine during that window, because starting a retinoid and a collagen powder in the same month tells you nothing about either. Then stop for a month. The 2014 trial found elasticity still elevated four weeks after discontinuation,[2] so a genuine effect should not vanish the week you run out, and anything that does was probably the water.
And keep the hierarchy straight: daily sunscreen, sleep, not smoking and enough total protein each do more for skin than any single scoop of powder. Among topicals, copper peptides are the other peptide with real mechanistic work behind them, and the injectable end of the peptide trend — BPC-157 and its relatives — has nothing like this evidence base.
Common questions
Does drinking collagen work?
Yes, modestly, and mainly for skin. Across two meta-analyses covering 19 and 26 randomized, double-blind, placebo-controlled trials, hydrolyzed collagen significantly improved skin hydration and elasticity compared with placebo, with smaller and less consistent effects on wrinkles. The honest summary is a small-to-moderate measurable change after 8 to 12 weeks of daily use, not the dramatic before-and-after in the marketing photos.
How much collagen should you take a day?
The trials that worked clustered between 2.5 and 10 grams of hydrolyzed collagen peptides a day, with 2.5 to 5 grams the most common effective dose for skin. Below about 1 gram a day the evidence thins out quickly. Taking more than 10 grams has not been shown to work better for skin; it mostly adds cost and calories.
How long does collagen take to work?
Plan on 8 to 12 weeks before judging it. Hydration tended to move first, often by week 6, with elasticity and wrinkle measures following between weeks 8 and 12. One trial still found a measurable elasticity benefit four weeks after participants stopped taking it, which suggests the change is in the skin rather than in the water you drank that morning.
Which collagen type is best for skin - I, II or III?
Type I with some type III. Human skin is built mostly from type I collagen with type III alongside it, so skin products use hydrolyzed type I and III from fish, cattle or pigs. Type II collagen is cartilage collagen and belongs to the joint conversation, where the undenatured form is used at around 40 mg a day through a completely different, immune-mediated mechanism. Buying a type II product for your face is a category error.
Is marine collagen better than bovine collagen?
There is no convincing evidence that it is. A meta-analysis of 26 trials found no significant difference between collagen sources for skin elasticity, and while source did shift the hydration results in subgroup analysis, the trials were not designed to compare sources head to head. Pick on the basis of allergies, diet, price and third-party purity testing rather than the marketing.
Does collagen build muscle?
No, and that is the one claim worth being blunt about. Collagen is a protein, but it is almost free of leucine, the amino acid that triggers muscle protein synthesis. In a randomized crossover trial in trained adults, an equal amount of protein from lactalbumin raised myofibrillar protein synthesis about 13 percent more than collagen peptides did. Count collagen as a skin supplement, not as part of your protein target.
Can you get collagen from food instead of a supplement?
Partly. Gelatin, bone broth, skin-on fish, chicken skin and slow-cooked cuts all contain collagen, but the dose in a bowl of home-made broth is variable and usually well below the 2.5 to 10 grams used in the trials. Collagen is also not an essential nutrient: your body builds it from glycine, proline and vitamin C, which is why total protein intake and a diet with enough vitamin C matter more than any single scoop.
Is the collagen skin research trustworthy?
Partly, and the caveat is structural rather than scandalous. A large share of collagen skin trials are funded or run by companies that make collagen ingredients, and funding source is consistently associated with more favorable conclusions in nutrition research. The designs themselves are good - randomized, double-blind, placebo-controlled - and independent pooled analyses agree the effect is real. Expect the true effect to sit at the smaller end of what the studies report.
Does collagen have side effects?
Reported side effects are minor and uncommon: mild bloating, a feeling of fullness, or an aftertaste. The trials report clean safety records at 2.5 to 10 grams a day for up to six months. The real risks are allergic reactions to the source material, fish in particular, and contamination, which is why third-party purity testing is worth paying for in an animal-derived powder.