About 9,300 searches a month ask about collagen pills — most of them "collagen pills", some "collagen pills vs powder". The form question has a short answer and a long one. The short answer: nobody has run the trial. The long answer is arithmetic, because the trials behind collagen's muscle, joint and tendon claims gave grams a day, and capsules hold milligrams. This page reports what the labels hold and what the trials used; it does not suggest a form or an amount (how we work).
What the trials gave
| Outcome | Trial | Amount a day |
|---|---|---|
| Muscle mass with training, older men | Zdzieblik 2015, Br J Nutr | 15 g |
| Muscle mass with training, women | Jendricke 2019, Nutrients | 15 g |
| Activity-related joint pain, athletes | Clark 2008, Curr Med Res Opin | 10 g |
| Knee pain, athletes | Zdzieblik 2017, Appl Physiol Nutr Metab | 5 g |
| Achilles tendon, with training | Jerger 2022, Scand J Med Sci Sports | 5 g |
| Bone density, postmenopausal women | König 2018, Nutrients | 5 g |
Funding, comparators and what each one measured are on the collagen dose-by-goal page and collagen peptides benefits. The 15 g muscle trials, for example, include one whose supplier relationship was disclosed only in a 2025 corrigendum.
What capsules and tablets hold
These counts are a sample. The NIH label register returned 1,156 on-market "collagen" products on 2026-10-11, and its hourly request limit stopped our reads at 759 full labels. Capsules, tablets and softgels are tallied from those 759 only; where one product comes in several counts or sizes, each listing is a separate row.
Of those, 212 are capsules (152), tablets (50) or softgels (10), from 127 brands. 176 state a collagen amount per serving and 174 an amount we could divide per unit.
| Measure | Value |
|---|---|
| Median collagen per capsule or tablet (174 labels) | 500 mg (interquartile range 500–750) |
| Units holding 1,000 mg or more | 19 |
| Most common units per serving | 3 (71 labels), then 2 (53), then 1 (47) |
| Median collagen per listed serving (176 labels) | 1.4 g (interquartile range 1.0–2.2 g) |
| Servings of at least 2.5 g | 43 of 176 |
| Servings of at least 5 g | 11 of 176 |
| Servings under 1 g | 34 of 176 |
The capsule count
The arithmetic follows from a 500 mg capsule:
| Trial amount | What it was used for | Capsules at 500 mg |
|---|---|---|
| 5 g | knee pain, Achilles tendon, bone density | 10 |
| 10 g | joint pain in athletes | 20 |
| 15 g | muscle mass with training | 30 |
No label we read directs anything close to that. The median serving, 1.4 g, is under a third of the smallest trial amount, and 165 of the 176 servings with a stated amount fall short of 5 g. That is not evidence that capsules fail — no trial has tested them — but it does mean a capsule label and a trial result usually describe different amounts.
Powders, for comparison
Powders in the same sample held about seven times as much per serving:
| Form | Labels with an amount | Median per serving |
|---|---|---|
| Powder | 403 | 10.3 g (interquartile range 7.8–15 g) |
| Capsule, tablet, softgel | 176 | 1.4 g |
290 of the 403 powders gave at least 10 g a serving. A separate count of collagen-peptide powder servings against the trial doses — 212 labels read on 2026-09-27 — found a median of 11 g, so the two snapshots agree. Ready-to-drink liquids sit in between, at a median of 3.3 g (liquid collagen).
A different product sold in the same aisle
Some collagen capsules are not a protein dose at all. Undenatured type II collagen (often sold as UC-II) is taken at around 40 mg a day for joint symptoms, on a proposed mechanism called oral tolerance, and its trials stand apart from the hydrolysed-collagen literature. A capsule label listing tens of milligrams of type II collagen is that product, not an underfilled version of a powder. The best-known trial gave 40 mg a day of UC-II against glucosamine–chondroitin and placebo in knee osteoarthritis (Lugo et al., Nutr J 2016, PMID 26822714); it is a joint-symptom trial in people with arthritis, not a muscle or tendon trial.
What has not been tested
- Absorption by form. A PubMed search on 2026-10-11 — collagen, capsules or tablets, powder, and absorption or bioavailability — returned one record, an animal study of a herbal capsule. No human study has compared plasma collagen peptides after capsules against powder.
- Outcome by form. No trial has given the same collagen at the same amount as capsules and as powder and compared results.
The creatine version of this question, with the same kind of label arithmetic, is on creatine pills vs powder. Ready-to-drink forms are on liquid collagen.
What this page cannot tell you
- A capsule's milligrams are what the panel declares. The register stores the printed panel and runs no assays, so an underfilled capsule would look the same here as an accurate one.
- Collagen amounts were read from the supplement-facts rows naming collagen or gelatin; blends that list collagen inside a proprietary total were counted only where an amount is stated.
- No figure here is a suggested amount or a reason to choose a form. Allergy and kidney-stone questions — the oxalate pathway and source-animal allergens — belong with a clinician.
Sources
- NIH Office of Dietary Supplements. Dietary Supplement Label Database — API v9: 1,156 on-market products named "collagen", 759 labels read on 2026-10-11, of which 212 capsules, tablets and softgels and 471 powders — dsld.od.nih.gov
- Lugo JP, Saiyed ZM, Lane NE. Efficacy and tolerability of an undenatured type II collagen supplement in modulating knee osteoarthritis symptoms. Nutr J 2016;15:14 — PMID 26822714
- Trial amounts as recorded on our collagen dose page: Zdzieblik 2015 (Br J Nutr), Jendricke 2019 (Nutrients), Clark 2008 (Curr Med Res Opin, PMID 18416885), Zdzieblik 2017 (Appl Physiol Nutr Metab, PMID 28177710), Jerger 2022 (Scand J Med Sci Sports, PMID 35403756), König 2018 (Nutrients).
- PubMed search, collagen with capsules or tablets, powder, and absorption or bioavailability, run 2026-10-11 (1 record).