Does the collagen I swallow end up in my skin?

Not as collagen. It is taken apart in the gut. What does survive is fragments: a two-unit piece called Pro-Hyp is measurable in blood within an hour of a 10 g drink and peaks between one and two hours. That much is settled, in groups of five and six people. Whether any of it then tells skin or cartilage to build more collagen has never been shown in a person — that step is an argument, not a measurement.

Why is there no health claim about skin on the packet in Europe?

Because a manufacturer asked for one and was refused. In 2013 the European Food Safety Authority assessed Gelita AG's application to link its collagen hydrolysate VeriSol P to a change in skin elasticity. It concluded that a cause and effect relationship had not been established. There is therefore no authorised health claim about skin or joints for collagen under the EU health claims regulation, and these products may not be marketed as having medical effects.

Does it matter whether it comes from fish, cattle or pigs?

For getting fragments into the blood, mostly not. A crossover in 6 healthy adults compared fish skin, pig skin and two sizes of cattle-hide product at 10 g each. The free hydroxyproline taken up was comparable across the sources. Total hydroxyproline in the blood did differ significantly between some of them, with the pig-skin product higher than the fish one. For the skin results, the pooled analysis of 26 trials found no significant difference between sources for springiness (p = 0.21), and reported no equivalent test for water content. For allergy it matters completely — fish collagen is a fish allergen in its own right.

Is there lead or arsenic in collagen powder?

One published analysis has looked, and it found less than the limits allow. Researchers measured arsenic, lead, chromium, cadmium and mercury in 120 samples of marine collagen — cod, pangasius, jellyfish, tilapia and mackerel skin — from four commercial brands plus one experimental extract, collected in 2022 and 2023. Arsenic was the most abundant, averaging 0.59 mg per kg and reaching 1.11 at most; lead averaged 0.138. No sample exceeded the maximum levels set by the European Union, and the estimated daily intakes stayed below the tolerable limits. That is one study of marine products from four brands, not a statement about any particular tub on a shelf.

Who paid for the studies?

Almost every individual trial named on this page was paid for or supplied by a collagen manufacturer. GELITA AG part-funded the 12-month bone trial. GELITA supplied the product for the 180-person knee trial and part-funded it. Gelita AG provided the samples for the 53-man muscle trial, and paid an amount covering participant compensation and sample analysis. In August 2025 that journal issued a correction adding the funding detail, and disclosing that a co-author is named on a patent application for the substance. The 6-person absorption study was funded by Rousselot BV, with several of its authors employed by the company. The two pooled analyses that gathered the skin trials together are the exception: one was funded by a Taipei hospital, the other reports no funding at all. Industry money does not make a result wrong. It means these particular results have not been reproduced by an independent group.

Are there finished trials whose results were never published?

Yes, and they are not small. A Shanghai supplement company ran a 90-person trial of 5 g a day against a dummy drink for 12 weeks, measuring skin firmness, thickness and density with instruments. It completed on 19 November 2024 and no results have been posted. A second study put 264 people on vegan, cattle, fish or chicken collagen at 2.5 g, 5 g or 10 g, measuring crow's feet wrinkles and hair thickness. It completed on 30 June 2023 and no results have been posted there either.

Supplement

Collagen peptides (hydrolysed collagen)

Also known as hydrolysed collagen · collagen hydrolysate · collagen peptides · gelatin hydrolysate

Tested in people, but barely

19 of 51 peptides sit at this level

Category: Supplement
Doping status: Not banned in sport
Sources: 20

Collagen peptides are a powder made by breaking down the skin, hide or bone of cattle, pigs or fish with enzymes, sold as a food supplement rather than a medicine. Twenty-six randomised trials in 1,721 adults have been pooled for skin: instruments pressed against the skin read more water and more springiness after 2 to 12 weeks. What almost nobody has tested is whether a person can tell. Beyond skin there is one 12-month bone trial, one disputed muscle trial and four knee trials judged at high risk of bias — and a collagen manufacturer paid for or supplied the product in most of them.

Collagen peptides (hydrolysed collagen)What it is

What it is
Sold and legal
In your body
What changed
What can go wrong
Amounts studied
What people say
Unknowns
Questions
Sources

What it is

Not a single peptide but a mixture of thousands of different fragments from collagen broken down with enzymes, sourced from the skin, bone or hide of cattle, pigs or fish. It is rich in glycine, proline and hydroxyproline, but the make-up varies between manufacturers and batches, so no single sequence can be given.

How it is sold, and whether it is legal

Sold as: Swallowed, as a powder or capsule.
Legal status in the EU: A food and food supplement in the EU, not a medicine. In 2013 the European Food Safety Authority rejected an application from Gelita AG to link the collagen hydrolysate VeriSol P to skin elasticity, concluding that a cause-and-effect relationship had not been established. There is therefore no authorised health claim about skin or joints under EU Regulation 1924/2006, and these products may not be marketed as having medical effects.

What it does in your body

6 parts of the body · 2 measured in people, 1 where studies in people disagree, 3 from one small study

Two mechanisms are proposed, neither proven in humans: that the peptides supply raw material for the body's own collagen production, or that they act as signals. The objection that collagen just gets broken down into individual amino acids is not entirely right — the two-unit fragment prolyl-hydroxyproline and related fragments have been measured intact in the blood after someone swallows them. But a peptide reaching the blood says nothing about whether it controls collagen production in skin or joint cartilage, and that link has not been shown in people.

The skin: Swallowed collagen does not travel to the skin as collagen. It is taken apart in the gut. What the trials measure is what happens weeks later, using a probe held against the forearm or the cheek. The probe reads two things: how much water the outer skin holds, and how quickly the skin springs back when it is sucked up slightly. Both read higher on collagen than on a dummy powder when the trials are pooled. The difference was larger in trials that ran longer than eight weeks than in shorter ones.; Pooled from 26 randomised trials against a dummy powder. Between them: 1,721 adults aged 21 to 70, mostly women, given between 0.6 g and 12 g a day for 2 to 12 weeks. A second pooled analysis sat inside a wider review of supplements for sun-aged skin. It covered 8 collagen comparisons in 306 people and pointed the same way on springiness, with a much wider spread. Against that, the European Food Safety Authority refused a skin-elasticity claim for one manufacturer's collagen hydrolysate in 2013. It concluded that a cause and effect relationship had not been established.; Measured in people; Source [04]Source [05]Source [06]
Blood, in the hours after you drink it: Within an hour of drinking it, fragments of the powder are measurable in the blood. The one that turns up in the largest amounts is a two-unit fragment called Pro-Hyp, and it peaks between one and two hours after the drink. This is the part of the story that is genuinely settled: some of it does survive digestion intact. What it does after that is not settled. A fragment in the blood is not the same thing as a signal arriving at skin or cartilage, and no study in a person has followed it that far.; A crossover study in 6 healthy adults, 3 women and 3 men aged 19 to 50. Each was given a single 10 g dose of four different collagen products a week apart, with blood taken for six hours. Earlier, a pilot in 5 healthy women given 5 g after a 12-hour fast, with blood taken over eight hours. In that pilot the straight-chain fragment peaked at about 3.87 nmol per ml at two hours, and the ring-shaped version at about 0.21. Both studies are tiny.; Measured in people; Source [03]Source [07]
Knees and other joints: In people whose knee cartilage has worn down, pain scores fell further on collagen than on a dummy powder when three trials were pooled. In young, active people with sore knees, the difference was there but small: about six millimetres more improvement on a 100-millimetre line where the person marks their own pain. Nothing in these trials shows the joint itself changing — no scan, no cartilage measurement, only what people reported and what a doctor scored.; Pain pooled from 3 randomised trials in 375 patients with knee osteoarthritis, given between 6.8 g and 10 g a day for 6 weeks to 6 months. All four trials in the review were judged at high overall risk of bias. The reviewers rated the strength of the pain evidence moderate and the strength of the side-effect evidence very low. Separately, one 12-week trial in 180 physically active people aged 18 to 30.; Studies in people disagree; Source [02]Source [08]
Bones: These were postmenopausal women whose bones had already thinned with age. A year of 5 g a day nudged bone density up in the lower spine and in the neck of the thigh bone. Over the same year, the same measurement fell in the women taking a dummy powder. Blood markers moved in the matching direction: the marker of bone being built rose on collagen, and the marker of bone being broken down rose only on the dummy powder.; One randomised trial against a dummy powder: 131 postmenopausal women enrolled and all 131 counted in the analysis, 102 completing, average age 64, 12 months. This is the longest published trial of collagen peptides in people. Part of the study was paid for by GELITA AG. Nobody has repeated it.; One small study; Source [09]
Muscle and body fat: One trial gave older men with age-related muscle loss 15 g a day while every man in the trial did the same supervised weight training three times a week. The collagen group gained more lean tissue, lost more fat and got stronger in the thigh than the men doing identical training on a dummy powder. The size of that difference is contested. A letter in the same journal argued the gain was larger than anything reported for protein supplements in this group, and larger than a testosterone trial had achieved. The same letter said the 15 g contained only 0.4 g of leucine — the amino acid that normally triggers muscle building — which it called far too little to have any effect.; One randomised trial against a dummy powder: 53 men, average age 72, 12 weeks, all training. Gelita AG provided the samples and paid an amount covering participant compensation and sample analysis. In August 2025, ten years after publication, the journal issued a correction. It added that funding statement and disclosed that a co-author is named on a patent application covering the substance.; One small study; Source [10]Source [11]Source [12]
Kidneys and urine: Collagen is unusually rich in hydroxyproline, an amino acid the body converts partly into oxalate — the substance that most kidney stones are made of. Healthy adults were fed a large daily dose of gelatin, which is the same protein less finely broken down. The amount of oxalate in their urine rose by about two-fifths, and a related substance rose more than fivefold. Nothing followed those people to see whether stones formed. The study was measuring the chemistry, not the outcome.; Ten healthy adults, five men and five women, average age 32, on two controlled five-day diets: one with 30 g of gelatin a day, one with whey protein instead. A separate dose-response test in 6 adults gave single loads of 1 g, 2 g, 5 g and 10 g of gelatin. The 5 g and 10 g loads moved urine chemistry; the 1 g and 2 g loads did not.; One small study; Source [13]

What changed when it was measured

8 findings · 2 measured in people, 1 where studies in people disagree, 5 from one small study

Miranda and colleagues (International Journal of Dermatology 2021) pooled 19 randomised double-blind studies with 1,125 participants, aged 20–70 and about 95 per cent women, and reported better skin hydration, elasticity and wrinkle appearance after around 90 days. For knee osteoarthritis there is a pooled analysis of only four randomised studies with 507 patients in total: pain relief with a standardised mean difference of −0.58 (95 per cent confidence interval −0.98 to −0.18), but all four were judged to be at high risk of bias. Absorption was examined in a pilot study of five healthy women (Nutrients 2018), where the straight-chain form reached at most about 3.87 nmol/ml in blood two hours after intake, while the ring form the study focused on sat at around 0.21 nmol/ml.

Fragments are measurable in the blood within an hour of drinking it, and peak between one and two hours. Everything else is slow. The pooled skin trials ran 2 to 12 weeks, and the pooled analysis found a larger difference in the trials longer than eight weeks than in the shorter ones. The knee trials ran 6 weeks to 6 months. The only bone result comes from a full 12 months, which is also the longest anyone has been given collagen peptides in a published trial. There is no published data at all on what happens in year two.

How much water the outer skin holds: Higher on collagen than on a dummy powder, by a pooled standardised difference of 0.63 (95% confidence interval 0.38 to 0.88). A standardised difference is a way of averaging results from machines that do not share a unit. A value of 0.63 is conventionally read as a moderate difference between the two groups, not as a number anyone can see in a mirror.; 26 randomised trials against a dummy powder, 1,721 adults aged 21 to 70, mostly women, 0.6 g to 12 g a day for 2 to 12 weeks; Measured in people; Source [05]
How quickly the skin springs back: Higher on collagen than on a dummy powder, by a pooled standardised difference of 0.72 (95% confidence interval 0.40 to 1.03). A separate pooled analysis of eight comparisons in 306 people reported 1.93 (0.65 to 3.20). That is a much larger figure with a much wider spread, which is what happens when a handful of small trials disagree with each other.; The same 26 trials in 1,721 adults; the second figure comes from 8 comparisons in 306 people inside a review of supplements for sun-aged skin; Measured in people; Source [05]Source [06]
Knee pain in worn-down knees: Lower on collagen than on a dummy powder, by a pooled standardised difference of −0.58 (95% confidence interval −0.98 to −0.18). The reviewers rated the strength of that evidence moderate. They also rated all four trials in the review at high overall risk of bias, meaning the way the trials were run leaves room for the result to be an artefact.; 375 patients with knee osteoarthritis across 3 randomised trials, 6.8 g to 10 g a day for 6 weeks to 6 months; Studies in people disagree; Source [02]
The volume of wrinkles at the corner of the eye: About 20% lower than on a dummy powder at four weeks and again at eight weeks, and still lower four weeks after stopping. In skin samples from some of the same women, the building block of new collagen read 65% higher and elastin 18% higher. A third protein, fibrillin, moved 6%, and that was not statistically significant — meaning it is within what chance produces.; 114 women aged 45 to 65, 2.5 g a day for 8 weeks, against a dummy powder; One small study; Source [14]
Knee pain during activity in young, active people: Fell by 21.9 mm on collagen against 15.6 mm on a dummy powder, on a 100-mm line the person marks themselves (p = 0.024). Scored by a doctor instead, 23.0 mm against 14.6 mm (p = 0.003). Pain at rest and pain after 20 squats showed no difference between the groups, which the authors put down to those scores already being near zero at the start.; 180 people aged 18 to 30 with exercise-related knee pain, 5 g a day for 12 weeks; One small study; Source [08]
Bone density in the spine and the neck of the thigh bone: Spine: the standard bone score rose 0.1 on collagen and fell 0.03 on a dummy powder (p = 0.030). Thigh bone neck: rose 0.09 against a fall of 0.01 (p = 0.003). The authors describe this as bone density rising by almost 3.0% in the spine and 6.7% in the thigh bone neck on collagen. On the dummy powder it fell 1.3% and 1.0% over the same year.; 131 postmenopausal women with age-related bone thinning, 5 g a day for 12 months; One small study; Source [09]
Lean tissue, body fat and leg strength in older men who were weight training: Lean tissue rose 4.2 kg against 2.9 kg on a dummy powder; fat fell 5.4 kg against 3.5 kg; thigh strength rose 16.5 against 7.3 units of turning force. All three differences were reported as statistically significant. A letter in the same journal argued the lean tissue gain is unmatched by any protein supplement study in this group. It also argued that the fat loss is about 80% of what much longer periods of dieting produce in older men, reached here in 12 weeks.; 53 men, average age 72, with age-related muscle loss, 15 g a day for 12 weeks, all doing the same supervised training; One small study; Source [10]Source [11]
Oxalate in the urine: 43% higher on five days of 30 g gelatin a day than on the same diet with whey protein instead; a related substance, glycolate, was 5.3 times higher. Single loads of 5 g and 10 g of gelatin moved both; loads of 1 g and 2 g did not.; 10 healthy adults, average age 32, on two controlled five-day diets; a dose-response test in 6 further adults; One small study; Source [13]

What can go wrong

5 effects, 1 serious

No serious safety signals have come out of the pooled studies, and the supplement is handled like any ordinary protein food. The real uncertainties are about method: the follow-up periods are short (often around 90 days), the studies small, the outcomes instrument readings rather than things patients notice, all four osteoarthritis studies are at high risk of bias, and the food safety authority judged the evidence insufficient. Anyone allergic to fish or shellfish should check what the product is made from.

A serious allergic reaction to the animal it was made fromSerious: Fish collagen is a fish allergen in its own right, separate from the protein usually blamed for fish allergy. A published case describes a factory worker who cut and packed fish collagen and shark cartilage powder. Ten months into the job they had a blocked nose, sore eyes and hives at work. Two years after those first symptoms they suddenly began reacting to eating fish, with hives all over, wheezing and streaming eyes. Tests for the protein usually blamed for fish allergy came back negative. The label is the only thing that says whether a tub is fish, cow or pig.; Not measured in any trial. In the US regulator's public database of food and supplement complaints, 8 of the 56 reports naming a product called collagen peptides used the term hypersensitivity. That is the database's word for an allergic reaction, and it does not by itself mean a severe one. It is the single most-used term in that group of reports.; Source [15]Source [16]
Stomach and gut upset — feeling sick, loose stools, being sick, stomach ache: This is the commonest complaint wherever anyone has asked. It is also the commonest complaint about any powder swallowed daily on an empty stomach, and the trials do not separate the two. Across the four pooled knee trials, side effects of every kind were reported by enough people in both groups that the difference could easily be chance.; In a survey of 56 people who had taken collagen, 32.1% reported gut symptoms. In the US complaints database, feeling sick, loose stools, being sick and stomach ache are all among the ten most-used terms for products named collagen peptides.; Source [15]Source [17]
Any side effect at all, counted against a dummy powder: The confidence interval just touches 1, which is the point at which a result stops being distinguishable from no difference. So this is a signal of a possible small excess, not a demonstration of one. The events recorded across those trials were stomach and gut complaints, migraine, chest infections, a joint infection and thigh pain. The 12-month bone trial and the 180-person knee trial each state that no adverse events were noted.; Odds of 1.66 against a dummy powder (95% confidence interval 0.99 to 2.78, p = 0.05) across 4 randomised trials in 507 knee patients. The reviewers rated the strength of that evidence very low.; Source [02]Source [08]Source [09]
Headache and migraine: Nobody has tested this against a dummy powder as a specific question. So there is no way to tell from these numbers whether the powder causes headaches, or whether people who get headaches sometimes blame the newest thing they started taking.; 7 of the 56 US database reports naming collagen peptides used the term headache and 4 used migraine. Migraine was also among the events recorded in the four pooled knee trials.; Source [02]Source [15]
More oxalate in the urine: Oxalate is what most kidney stones are made of, and collagen is unusually rich in the amino acid the body turns into it. Single loads of 5 g and 10 g of gelatin were enough to move it; 1 g and 2 g were not. No study has followed anyone long enough to see whether stones actually form, and no study has done this in people who already make stones.; Measured, not counted: 43% more oxalate in the urine of 10 healthy adults over five days of 30 g gelatin a day than over five days of whey protein.; Source [13]

Who it is known to be dangerous for

Two groups are established. The first is anyone allergic to the animal the powder is made from. Fish collagen is itself a fish allergen, separate from the protein usually blamed for fish allergy, and a tub labelled only "collagen peptides" does not say what it came from. The second is anyone who forms calcium oxalate kidney stones, or who has the inherited condition that overproduces oxalate. Collagen is rich in the amino acid the body converts into oxalate, and a daily 30 g gelatin load raised urinary oxalate by 43% in healthy adults within five days. Beyond those two, nothing has been established, and the reason matters. This is regulated as a food, so it has never been given the warnings section that an approved medicine gets. Every trial described here was in adults. Nothing has been published on pregnancy, on breastfeeding, on children, or on whether it interacts with any medicine.

The amounts the studies used

8 amounts

These are the amounts the studies below gave their participants, and they are here as facts about those studies. They are not a recommendation, not a starting point and not a range to pick from. What a person should take, if anything, is a question for a doctor who knows them.

Proksch et al. 2014, wrinkles at the corner of the eye in women aged 45 to 65: 2.5 g of a collagen hydrolysate once a day, as a drink, against a dummy powder; 8 weeks, with a further 4 weeks of follow-up after stopping; Source [14]
König et al. 2018, bone density in postmenopausal women — the longest published trial: 5 g a day, against a dummy powder; 12 months; Source [09]
Zdzieblik et al. 2021, knee pain during activity in people aged 18 to 30: 5 g a day, dissolved in water, against a dummy powder; 12 weeks; Source [08]
The four randomised trials of knee osteoarthritis that have been pooled. Each used one fixed amount; the amounts differed between trials: Between 6.8 g and 10 g a day, from cattle, from chicken breastbone cartilage, or as a general collagen hydrolysate; 6 weeks to 6 months, depending on the trial; Source [02]
Zdzieblik et al. 2015, lean tissue and strength in older men with age-related muscle loss: 15 g a day, against a dummy powder, alongside three supervised weight-training sessions a week that both groups did; 12 weeks; Source [10]
The 26 randomised skin trials that have been pooled together. Each used one fixed amount; the amounts differed between trials: Between 0.6 g and 12 g a day across the trials, from fish in 14 of them, pig in 2, cattle in 1, chicken in 1 and an unstated source in 9; 2 to 12 weeks, depending on the trial; Source [05]
The absorption crossover in 6 healthy adults, measuring what reaches the blood: A single 10 g dose dissolved in 200 ml of water, of each of four products — fish skin, pig skin, and cattle hide in two fragment sizes — one week apart; A single dose each time, with blood taken for six hours afterwards; Source [07]
Knight et al. 2006, what hydroxyproline does to urine chemistry: 30 g of gelatin a day as part of a controlled diet, compared with whey protein; separately, single loads of 1 g, 2 g, 5 g and 10 g of gelatin; Two five-day diet periods; the single loads were followed for 24 hours; Source [13]

What people report

What follows is what people say online. It is not evidence, it is not graded, and it is not checked by anyone. People who had a bad time and people who are selling something both post more than people for whom nothing happened. It is here because you would go and read it anyway, and knowing who is talking is better than not.

Nobody can tell whether it is doing anything

The most common shape of a collagen thread is a person asking whether it works, and two or three people answering that they genuinely do not know. One poster had started taking it after tendon injuries, and wrote that their hair, skin and joints felt exactly the same. Another said hair and nails had clearly improved but could not say anything about their aches. A third had been taking it a month and wanted to know how long they were supposed to wait.

Read in Patient.info Communities, the thread "Does anyone take Collagen Peptides", 27–28 February 2020, four posts