Collagen for Women: Skin, Bone and Joint Doses From Trials

By Erin Rose · Published September 16, 2026 · Reviewed against primary sources · Methodology · About Us

Informational summary of published trials and of the labels manufacturers file with the NIH, not medical advice. If you are pregnant, breastfeeding, or being treated for osteoporosis or joint disease, the clinician managing your care decides.

Already know the dose you want? See the ranked collagen picks →

The dose table for women, by goal

Three things in that table decide a purchase: the doses are grams, not milligrams, except for the one joint product that is a different molecule; the trials used plain hydrolyzed peptides, not a named type; and every duration is two months or longer. The dosage guide has the goal finder; this page covers what the trials in women showed and what the women’s shelf sells.

Skin: the trials were run in women, at 2.5 grams

Almost every skin trial recruited women, so this is the one collagen claim where “for women” is not a marketing overlay. In the trial that set the dose, 69 women aged 35 to 55 took 2.5 g or 5 g of collagen hydrolysate or placebo daily for 8 weeks. Elasticity improved significantly against placebo in both collagen groups, and in the older women the gain was still measurable 4 weeks after they stopped; moisture moved the right way without reaching significance (Proksch 2014, PMID 23949208). The same group then gave 2.5 g a day to 114 women aged 45 to 65 for 8 weeks: eye-wrinkle volume fell about 20% against placebo and skin biopsies showed 65% more procollagen type I (Proksch 2014, PMID 24401291). A third trial gave 72 women aged 35 and over a drink carrying 2.5 g of peptides with vitamin C, zinc, biotin and vitamin E for 12 weeks and found hydration, elasticity, roughness and density all improved; because the drink was a blend, it cannot say what the collagen did alone (Bolke 2019, PMID 31627309).

The pooled evidence agrees with the single trials and adds the caveats. A 2021 meta-analysis of 19 randomized, double-blind trials with 1,125 participants aged 20 to 70, 95% of them women, found hydrolyzed collagen improved hydration, elasticity and wrinkles against placebo and put the effective course at about 90 days (de Miranda 2021, PMID 33742704). A 2023 meta-analysis of 26 trials with 1,721 participants confirmed the hydration and elasticity effects, found no significant difference between collagen sources on elasticity, and flagged bias in several of the included trials (Pu 2023, PMID 37432180).

As a buyer: the dose is 2.5 g, the product is a hydrolyzed peptide, the course is 8 to 12 weeks, and the effect is one a cutometer measures rather than one your friends notice. The Proksch trials were run with the manufacturer of the peptide they tested, which is normal in this category and the reason to treat effect sizes as a ceiling. Our collagen-for-skin page ranks products for this goal.

Bone after menopause: one good trial, at 5 grams for a year

Estrogen loss accelerates bone loss, which is why “collagen for bone” is pitched at women over 50. Type I collagen is the protein scaffold of bone, so the idea is not absurd. The evidence is one trial. In a randomized, double-blind study, 131 postmenopausal women with age-related low bone density (mean age 64, average spine T-score −2.4, in the osteopenia-to-osteoporosis range) took 5 g a day of specific collagen peptides or placebo for 12 months. Spine T-score rose 0.1 on collagen against a 0.03 fall on placebo; femoral-neck T-score rose 0.09 against a 0.01 fall. The bone-formation marker P1NP rose in the collagen group and the bone-resorption marker CTX rose in the placebo group. 102 of the 131 women completed the year (König 2018, PMID 29337906).

Two later pieces of evidence pull in opposite directions. Thirty-one of those women kept taking the peptides in an open-label follow-up to a total of four years and showed a continued rise in spine and femoral-neck density, but with no control group the number tells you what happened, not what the collagen did (Zdzieblik 2021, PMID 34520654). And a 2025 randomized trial in menopausal women gave 5 g of fish collagen peptides a day, with or without 1,000 mg calcium and 400 IU vitamin D, for six months: none of the bone-turnover markers changed in any group, while skin elasticity improved about 12% on collagen alone (Duangjai 2025, PMID 41002783). Six months may simply be too short for bone, and the 12-month trial measured density rather than markers, but the honest summary is one positive trial from a manufacturer-affiliated group, not a body of evidence.

In practice: if you are postmenopausal and want to try it, the tested dose is 5 g a day for a year, in a powder, on top of the things with fracture data behind them. Collagen has none. Adequate calcium (the calcium dosage guide), a vitamin D level that is not deficient (the vitamin D for women guide covers the fracture trials), resistance exercise and a bone-density scan when your clinician recommends one come first.

Joints: 10 grams of peptides, or 40 milligrams of a different product

Joint pain is where the collagen aisle splits into two products that share a name. The gram-scale product is the same hydrolyzed peptide as the skin and bone trials. In the trial usually cited, 147 college athletes with activity-related joint pain and no joint disease, 75 of them women, took 10 g of collagen hydrolysate or placebo daily for 24 weeks; the collagen group reported less pain at rest, walking, standing, carrying and lifting. Only 97 of the 147 could be analysed, a large hole for a trial of this size (Clark 2008, PMID 18416885). In diagnosed osteoarthritis, a 2019 meta-analysis found collagen lowered the total WOMAC score and stiffness, and pain on a visual analogue scale, but not the WOMAC pain or function subscales (García-Coronado 2019, PMID 30368550); a 2021 critique of that meta-analysis concluded its results do not support strong conclusions (von Hippel 2021, PMID 34636929); and a 2023 meta-analysis of 4 knee-osteoarthritis trials in 507 patients found a moderate reduction in pain while rating every included trial at high risk of bias (Lin 2023, PMID 37717022).

The milligram-scale product is undenatured Type II collagen (UC-II), from chicken cartilage, kept in its native folded shape. It works as an immune signal rather than a building block, which is why the dose is 40 mg. In 191 people with knee osteoarthritis, 40 mg a day for 180 days beat both placebo and glucosamine plus chondroitin on the total WOMAC score, with no safety difference between groups (Lugo 2016, PMID 26822714). So a 10 g tub of peptides does not deliver the UC-II effect, and a 40 mg UC-II capsule does nothing for skin or bone. On our census, products sold as undenatured Type II direct a median of 40 mg a day, the trial dose, while products selling hydrolyzed “Type II” direct a median of 2 g, a dose with no knee trial behind it. None of our ranked products is a UC-II; for that one look for “undenatured” and 40 mg on the panel, and see the types-compared guide.

Two more claims on the women’s shelf: muscle holds up with a condition, hair does not

The muscle trial was run in premenopausal women. Seventy-seven women did resistance training three times a week for 12 weeks and took 15 g of collagen peptides or placebo daily; the collagen group gained more fat-free mass and hand-grip strength than the training-only group (Jendricke 2019, PMID 31010031). The condition is the training: there is no trial showing collagen builds muscle in women who are not lifting, and it is an incomplete protein, so keep whey or another complete protein as the protein source.

Hair is the weakest claim on the shelf and the one most often printed on a tub sold to women. No trial has tested collagen alone with hair as its main outcome. The one placebo-controlled trial with a collagen-alone arm, the 5 g fish-collagen arm of the Duangjai 2025 bone trial cited above, reported less hair shedding than placebo as a secondary, self-reported outcome, which is a lead worth a proper trial, not a result. The one randomized hair-loss trial tested a tablet of 300 mg fish collagen with taurine, cysteine, methionine, iron and selenium, added to prescription treatment, in 83 people; the supplement arm scored better on photographs, the trial had no placebo, and the authors were employees of the company selling it (Milani 2023, PMID 37357646). Nails have one open-label trial: 25 participants, 2.5 g a day for 24 weeks, 12% faster growth and 42% fewer broken nails, no placebo (Hexsel 2017, PMID 28786550). If hair is the reason you are buying, the evidence does not support the purchase.

We read 746 collagen labels: what the women’s shelf carries

Original research Every on-market collagen label filed with the NIH Dietary Supplement Label Database that we could read (46,040 collagen-containing labels on the market, 851 read, 746 on the general shelf, 611 carrying a parseable collagen dose), classified by the type, form and source the label declares, whether it adds vitamin C or hyaluronic acid, and the daily dose its own directions add up to.

There is no “women’s” population in this census: the NIH database does not tag labels by intended audience, and we did not invent one from product names, so the figures describe the whole collagen shelf. What it sells is a set of distinctions the trials never tested. 43.2% of labels declare a collagen type at all. Of those that do, 86.3% list Type I, 77.3% Type III and 44.7% Type II. 12.3% of the shelf sells a “multi-collagen” blend of three or more types. None of the trials above selected a product by type: the skin and bone trials used hydrolysed collagen from animal skin or fish, which is Type I (with Type III when bovine) because of where it comes from, not because a type was chosen.

Vitamin C is the add-in with a mechanism behind it. 30% of labels co-formulate it. Your body needs vitamin C as a cofactor for the enzymes that stabilise the collagen triple helix; the NIH puts the adult female RDA at 75 mg a day, 85 mg in pregnancy and 120 mg when breastfeeding (NIH ODS; Pullar 2017, PMID 28805671). The Bolke trial paired the two; the Proksch trials got their result from peptides alone; none compared collagen with and without vitamin C. A diet with fruit and vegetables covers the RDA, so a co-formulated label is fine but not worth a premium. Hyaluronic acid, on 17.4% of labels, is a separate ingredient and a separate purchase decision; its presence in a collagen tub is not a reason to pay more for the collagen.

Form and source. 84% of labels describe themselves as hydrolyzed or as peptides, the form every trial on this page used; 14.2% say neither. Of labels naming a source, 72% name bovine and 61% marine or fish. The source changes the dose more than the result: bovine-only labels direct a median of 11 g a day and marine-only labels 5 g, so a marine product sold “for women” often lands at the skin dose and under the bone dose, while a bovine scoop covers every dose in the table. Among the figures that did not hold in both halves of the sample, and so are not printed here, are the dose of multi-collagen blends and the make-up of those blends.

Method: 851 collagen labels retrieved from the NIH Dietary Supplement Label Database (DSLD) v9 on 2026-09-14, in two populations (the general collagen shelf, 746, and every on-market label matching “type II collagen”, 105); 611 carried a parseable collagen dose. Type, source, form and add-ins are read from the product name, the label’s own statements and the Supplement Facts ingredient names; daily dose is the collagen row multiplied by the maximum daily servings the label directs. Shares are stated against the labels that declared the attribute, never against labels read. Every figure was recomputed in both halves of the read order: 22 of 26 findings held, and the 4 that did not are not published as numbers. The census is open data (CC BY 4.0) at /collagen/types-compared.json. What it cannot tell you: which products women buy, whether a label’s figure matches the tub, or anything about products not filed with the NIH.

Powder vs capsule: whether the format can hold a women’s dose at all

Capsules look like a supplement and a scoop looks like a food; the trials used the food-sized amount. In our separate format census of 465 collagen labels, the median powder directs 11 g a day, above every dose in the table, and 3.1% fall below the 2.5 g skin dose. The median capsule or tablet label directs 1.8 g a day, and 70.8% direct less than 2.5 g even when you take every capsule the label allows.

The capsule gap is arithmetic, not a brand failing. At the median capsule strength, reaching the 10 g joint dose takes 20 capsules a day, and only 4 of the 144 capsule labels we read direct as many as twelve. Even the 2.5 g skin dose is 5 capsules at that strength. Liquids sit in between: on the 64 liquid labels in our third census the median serving is 10 g and 18.9% direct under 2.5 g a day; the liquid vs powder guide prices them per gram. The one capsule that is correctly small is UC-II, because 40 mg is its dose. For every gram-scale goal here, the format decision comes before the brand decision.

Method: 598 labels retrieved from the NIH Dietary Supplement Label Database, 465 sold as a powder or a capsule/tablet with a readable dose; daily dose is the Supplement Facts panel figure multiplied by the maximum daily servings the label directs. Every figure was recomputed in each half of the read order and held; liquid figures come from a separate census of 64 labels and only its findings that held are printed.

The product links below go to Amazon and we may earn a commission if you buy. It never changes which product we pick or what we say about it. How we choose.

Frequently asked questions

Related guides

• Best collagen overall: every ranked pick, by dose, type and cost per day
• Collagen dosage guide: the goal finder, and why UC-II is dosed in milligrams
• Collagen types compared: I vs II vs III, marine vs bovine, and the full label census
• Liquid collagen vs powder: what a liquid serving delivers and what it costs per gram
• Best collagen for skin: products ranked for the best-evidenced goal
• Vitamin D for women: pregnancy, breastfeeding and what the fracture trials found
• Calcium dosage guide: the target for women over 50 and how to split it
• Vitamin C dosage guide: the collagen-synthesis cofactor

Sources