Recovery is the softest word in the peptide aisle. It covers three questions that have almost nothing to do with each other — a torn tendon, the soreness after a hard session, and healing from surgery — and a single product page will answer all three at once. This page separates them, counts the human research that exists for each, and says plainly which compounds in this category have any human evidence at all. It is research journalism about research compounds: it reports what sources state and contains no usage guidance, per our editorial standards.
Three questions wearing one word
Structural injury. A tendon, ligament or muscle is damaged and has to remodel. The relevant endpoint is healed tissue and a return to training, measured months out.
Exercise-induced muscle damage. Nothing is torn. Hard eccentric work produces soreness, temporary strength loss and raised creatine kinase, and it resolves in days. The endpoint is how fast function comes back.
Post-surgical healing. A controlled wound with a known timeline, where the outcomes that matter are infection, wound closure and rehabilitation milestones.
A compound can have evidence for one of these and none for the others. In practice, marketing copy tends to borrow the strongest-sounding claim from any of the three and apply it to all.
The count, done on 2026-09-06
We searched PubMed tonight. The counts are the finding.
| Search | Records |
|---|---|
| BPC-157, with the clinical-trial or randomised-controlled-trial publication filter | 0 |
| TB-500 or TB500, all designs, all species | 31 |
| TB-500, restricted to human-indexed records | 16 (none a trial of TB-500) |
| BPC-157, TB-500, CJC-1295, ipamorelin, sermorelin or IGF-1 LR3, crossed with tendon, ligament, rotator cuff, muscle injury or muscle strain | 31 |
| The same six compounds crossed with tendon, ligament or rotator cuff, human-indexed only | 11 |
| Collagen crossed with tendon or ligament, randomised-controlled-trial filter | 111 |
The last two rows are the whole picture in miniature. The compounds sold in vials for connective-tissue recovery have eleven human-indexed records between them, none of them a randomised trial of recovery; the protein sold in tubs has 111 randomised trials.
What the injectable evidence actually is
The single human study of BPC-157 in a musculoskeletal setting is a retrospective series: twelve patients given intra-articular knee injections for unspecified chronic knee pain, of whom seven reported relief for more than six months. It appears in the 2025 HSS Journal systematic review, which screened the BPC-157 musculoskeletal literature and found 36 studies — "35 preclinical studies" and "1 clinical study". The reviewers' own conclusion is the one to quote: "The lack of clinical outcome and safety data prevents us from proposing evidence-based practice use guidelines", and "clinicians and athletes should exercise caution when considering the use of BPC-157" (HSS Journal 2025, PMC12313605).
TB-500 has no human trial of any design. It is not thymosin beta-4, either, but Ac-LKKTETQ, a seven-residue acetylated fragment of the 43-amino-acid protein's actin-binding site; the 2024 paper that built a urine assay for it states that "the biological effects of TB-500, however, have not been documented" (Rahaman et al., Journal of Chromatography B 2024). The full protein does have human trials — intravenous phase 1 studies aimed at cardiac ischaemia, and a nine-patient phase 2 of eye drops for severe dry eye — none of them about tendon, muscle or sport. MuscleLedger's page on the BPC-157 and TB-500 combination works through that literature study by study.
The growth-hormone secretagogues are sometimes sold for recovery on the theory that raising IGF-1 helps tissue repair. Their human record is thin in a different way: they raise the hormone reliably and were never tested on a healing endpoint. CJC-1295 produced dose-dependent 2- to 10-fold rises in plasma growth hormone in healthy adults over 28- and 49-day dosing studies (Teichman et al., JCEM 2006) — a pharmacology result, not a recovery result. Ipamorelin's only randomised human trial was in postoperative ileus after bowel resection, and it missed its endpoint: median time to first tolerated meal 25.3 hours on drug versus 32.6 on placebo, p = 0.15, in 114 analysable patients (Beck et al., International Journal of Colorectal Disease 2014). That is the closest thing to a healing trial the class has, and it is a null result in a hospital setting.
Two 2026 reviews reach the same place from opposite directions. A scoping review in the American Journal of Sports Medicine found that "67% of identified publications utilized preclinical animal models" and that human studies "were limited to a handful of investigations, most lacking robust controls or rigorous study designs" (Tewari et al., AJSM 2026). A Sports Medicine review by Mendias and Awan puts the market alongside the evidence: "Many unapproved peptides demonstrate favorable tissue repair and metabolic outcomes in animal models, but rigorous human safety data are scarce, and there is potential for serious harm to patients", in "a parallel 'gray market' of unapproved compounds… operating largely outside of regulatory oversight" (Mendias and Awan, Sports Medicine 2026).
What does have human evidence
The honest answer to "what helps connective tissue recover" points away from the vial.
Gelatin and collagen, taken before loading. The study everyone cites is Shaw and colleagues, 2017: eight healthy men in a randomised, double-blinded crossover took 5 or 15 g of vitamin C-enriched gelatin or placebo, then skipped rope for six minutes an hour later. Circulating glycine, proline, hydroxyproline and hydroxylysine peaked an hour after the drink, and the 15 g condition "showed double the amino-terminal propeptide of collagen I in their blood, indicating increased collagen synthesis" (Shaw et al., AJCN 2017, PMC5183725). Read it precisely: eight participants, all male, a blood marker of synthesis rather than a healed tendon, and an engineered-ligament model for the mechanical part. It is the best-designed human study in this corner of the field and it is still small.
Creatine, for the soreness question rather than the injury question. A 2021 systematic review and meta-analysis of randomised trials in the Journal of Food Biochemistry pooled the creatine literature on recovery following exercise-induced muscle damage (Jiaming and Rahimi 2021). Effects across this literature are modest and the outcome measures vary, but the trials exist, in people, with placebo groups. MuscleLedger's head-to-head on creatine and the research compounds sets out how far that evidence actually reaches, including the parts of it funded by the creatine industry.
Everything that is not a supplement. Progressive loading, sleep and adequate protein have larger and better-replicated effects on recovery than anything discussed above, and no vendor sells them.
Post-surgery and the rotator cuff
Two of the fastest-rising searches in this cluster are about surgical recovery and rotator-cuff repair. There is no randomised human trial of any research peptide in either setting. The rotator-cuff literature that does exist for these compounds is rat and rabbit tendon-to-bone healing work; the surgical literature is the ipamorelin ileus trial described above, which was about gut motility and missed its primary endpoint. Anyone offering a peptide protocol for a post-operative shoulder is extrapolating across a species boundary, a tissue boundary and a study-design boundary at once.
Banned status, briefly
BPC-157 is named under S0 (non-approved substances) on the WADA 2026 Prohibited List and thymosin-beta-4 and its derivatives, including TB-500, under S2.3 (growth factors); the growth-hormone secretagogues sit at S2.2. All are prohibited at all times, in and out of competition, and the NCAA's 2026-27 banned classes name the same compounds. Collagen, gelatin and creatine are not prohibited. Our page on peptides and drug testing explains which tests actually look for these molecules and which do not — the two questions are separate, and the answer to "will it be caught" has no bearing on whether it is banned.
What would move these rows
A randomised controlled trial in humans, with an imaging or return-to-activity endpoint, a control group, and enough participants to detect a difference. For BPC-157 that number currently stands at zero, and a 2025 commentary in Pharmaceuticals notes that the only Phase I trial it could identify was cancelled and that over 80% of BPC-157 publications originate from a single research group (Pharmaceuticals 2025, PMC12567171). Until such a trial exists, our evidence ledger scores these compounds 1 out of 5, and the score reflects the absence of data rather than a demonstrated failure. Those are different things, and both are reasons to be careful.
Sources and dates
- PubMed searches run 2026-09-06 (counts in the table above are reproducible with the terms as written).
- HSS Journal 2025 — BPC-157 musculoskeletal systematic review (PMC12313605)
- Tewari et al., American Journal of Sports Medicine 2026 — scoping review
- Mendias and Awan, Sports Medicine 2026
- Rahaman et al., Journal of Chromatography B 2024 — TB-500 urinary assay
- Shaw et al., American Journal of Clinical Nutrition 2017 — vitamin C-enriched gelatin (PMC5183725)
- Beck et al., International Journal of Colorectal Disease 2014 — ipamorelin in postoperative ileus
- Teichman et al., Journal of Clinical Endocrinology & Metabolism 2006 — CJC-1295
- Jiaming and Rahimi, Journal of Food Biochemistry 2021 — creatine and exercise-induced muscle damage
- Pharmaceuticals 2025 — BPC-157 commentary (PMC12567171)
- WADA 2026 Prohibited List, sections S0, S2.2 and S2.3; NCAA 2026-27 banned-drug classes.