The entire clinical case for taking collagen peptides to heal a torn tendon or ligament comes down to six human studies. Only one of them tested an actual tear using a clearly defined peptide.
That is the finding of a systematic review ↗ published August 14 in Frontiers in Medicine, which searched PubMed and a mirrored Scopus strategy through March 17, 2026 for human studies of hydrolyzed collagen peptides in tendon and ligament injuries. Hydrolyzed collagen, sold as collagen peptides in tubs of flavorless powder, is collagen protein chopped into short fragments small enough to absorb. It is one of the best-selling supplements in the world, marketed for skin, joints, and connective tissue. The question the reviewers asked is narrower and more useful than the marketing: does it help a tendon or ligament that has actually torn?
Six studies met the bar. Three looked directly at tears or surgical repair: a pilot study of injectable low-molecular-weight peptides in partial tears of the supraspinatus (a rotator cuff tendon in the shoulder), a randomized trial after ACL reconstruction in the knee, and a randomized trial after rotator cuff repair. The other three were only indirectly relevant, covering Achilles tendinopathy, chronic ankle instability, and rotator cuff tendinopathy, conditions where the tissue is irritated or degenerating rather than torn through.
Across those six, the direction was gently encouraging. Collagen-based treatment was linked to less pain, better function, lower painkiller use, or improved imaging. Taken at face value, that is a supplement doing something.
The face value does not survive contact with the details, and the reviewers say so. Most of the oral products in these studies were multicomponent, meaning the collagen came bundled with vitamin C, other amino acids, or additional ingredients, so any benefit cannot be pinned on the collagen itself. Only one of the direct-tear studies evaluated a clearly defined low-molecular-weight peptide formulation. The rest leave you unable to say what actually did the work, or whether anything did.
The reviewers' own summary is cautious to the point of being a warning label. The evidence, they write, is "sparse, heterogeneous, and low certainty." Heterogeneous means the studies disagreed with each other. Low certainty means the kind of evidence that shifts when the next real trial arrives. They land on a cautiously positive signal for collagen peptides as an add-on to rehabilitation or post-surgery care, and a call for blinded, placebo-controlled trials that use well-characterized formulations and measure outcomes patients actually care about.
The formulation problem is the one worth sitting with, because it is where the supplement aisle and the molecular world diverge. A collagen peptide on a supplement label is not a single molecule. It is an undefined mixture of fragments from boiled-down animal collagen, and two tubs from two brands are not the same product. That is the opposite of how a drug, or a peptide card on this platform, is defined, where the sequence is known and fixed. When the input is a mixture nobody has fully characterized, a clean answer about what it does to a healing tendon is hard to get even in principle.
None of this makes collagen peptides useless. It makes the honest state of the evidence thin, and thin in a specific way: mostly indirect, mostly multicomponent, and resting on one clean study of an actual tear. That is a starting point for a real trial, not a reason to expect a torn ACL to knit faster because of a scoop of powder.
peptidemodel tracks connective-tissue work under tissue repair ↗. It is the same skeptical spine as an earlier readout on the research peptides TB-500 and BPC-157, where TB-500 helped rat tendons heal and stacking BPC-157 added nothing ↗. Different molecules, same lesson: the marketing for connective-tissue peptides runs well ahead of the trials.