BPC-157: What the Evidence Actually Shows

Search BPC-157 and you get two internets. One tells you it is a wonder healing peptide that fixed someone's tendon in three weeks. The other tells you it is an unapproved research chemical with no human evidence and you should not touch it. Both are repeating a headline. Neither has looked at what actually happened to this compound, which is more interesting than either version.

We went through the clinical trial registries, the patent record, the FDA's own advisory committee documents, 227 published papers, third-party product assays, and about fifteen thousand posts across a decade of peptide forums. This is what we found. Some of it is more encouraging than the sceptical account allows. Some of it is considerably worse than the enthusiast account admits. We have tried to report both, say where each claim comes from, and let you weigh it yourself.

This part is unambiguous and it does not move. BPC-157 has no marketing authorisation in Ireland or anywhere in the European Union. Under Irish law a substance given to a person to restore, correct or modify physiological function is a medicinal product, and the Medicinal Products Regulations 2007 prohibit placing one on the market without an authorisation. Selling it, supplying it, importing it commercially and advertising it are unlawful in Ireland.

You will often read that BPC-157 is not a controlled substance. That is true, it appears nowhere in the Misuse of Drugs Regulations, but the sentence is usually doing work it should not. It means personal possession is not itself a criminal offence. It says nothing about supply. The HPRA detained over 750,000 units of illegal medicines across nearly 14,000 consignments in 2025, up 180 per cent on the year before, and packages from overseas peptide sellers are among what customs intercepts.

That is the position, and nothing further down this page changes it.

No human trials is technically true today, but it implies the compound was tested and found wanting. It was not.

In the early 2000s the Croatian pharmaceutical company Pliva developed BPC-157 as a drug candidate under the code PL-14736, for ulcerative colitis. This was not a token effort. They completed a full GLP toxicology package, four-week intravenous studies in rats and dogs and fourteen-day intracolonic studies in both species, published in Toxicology Letters in 2003. They ran a Phase 1 in healthy volunteers. And in 2005 they presented, at Digestive Disease Week, a multicentre, randomised, double-blind, placebo-controlled Phase 2 in mild-to-moderate ulcerative colitis. Fifty-three patients randomised, forty-six completed. The result trended in the drug's favour, a 3.2-point improvement in disease activity against 1.6 on placebo.

Then the programme stopped. Not because it failed. In May 2006, twelve months after that readout, Pliva sold its entire research institute, the building and all 130 employees, to GlaxoSmithKline for thirty-five million dollars, and turned away from branded pharmaceuticals toward generics. Five months later Barr Pharmaceuticals bought Pliva for 2.5 billion dollars. Teva bought Barr in 2008. A pentadecapeptide in Phase 2 for inflammatory bowel disease has no home inside a generics manufacturer.

We could not find a single document anywhere stating that PL-14736 was discontinued for lack of efficacy or for a safety signal, and we looked specifically for one. The programme appears to have died in a corporate restructuring.

The arithmetic is brutal and worth spelling out, because it explains a great deal that gets attributed to conspiracy.

The original composition-of-matter patent on BPC-157 expired in 2018. The molecule is public domain. A placebo-controlled pivotal trial costs on the order of thirty-five million dollars, and a compound entering Phase 1 has roughly a fourteen per cent chance of reaching approval, about four per cent for inflammatory indications. Against that, a sponsor could expect five years of US data exclusivity and a formulation patent running to 2033.

And uniquely, an approved BPC-157 product would launch into competition with an unregulated worldwide supply of the identical molecule, sold cheaply as a research chemical. That is not a market failure caused by anyone suppressing anything. It is a rational commercial no-go, and the peer-reviewed literature says so out loud. One 2026 review notes that the absence of exclusivity is a significant disincentive for industry-sponsored Phase II and III development, and may partly explain the paradox of extensive preclinical work alongside almost no clinical development.

So the enthusiast account has a real point. But it usually stops here, and the next part is the reason a sponsor still would not touch this even if the money were there.

Of the 227 papers indexed on PubMed for BPC-157, 175, seventy-seven per cent, share a single author: Predrag Sikiric, at the University of Zagreb. Seventy-eight per cent carry a Zagreb affiliation. That is an extraordinary concentration for a compound with thirty years of literature.

Virtually every one of those studies uses a single dose level, typically ten micrograms per kilogram. After three decades there is no dose-response curve, no minimum effective concentration, no maximum tolerated dose and no therapeutic window for BPC-157 in any species. There is no human pharmacokinetic data at all. Oral bioavailability has never been measured in any animal. In Pliva's Phase 1, the peptide was not detectable in plasma after rectal dosing.

There is also a conflict-of-interest problem, reported by STAT News and Undark in February 2026 and verifiable against the patent registers. Sikiric is named on BPC-157 patent applications going back to at least 1989. He is listed as an owner of PharmaCotherapia, the company that sponsored the only human trial ever registered on ClinicalTrials.gov. He is listed as chief executive of Diagen, which holds the live formulation patent on a stable BPC-157 salt, a patent the company currently advertises for sale on its website. None of these interests were disclosed in the papers the reporters reviewed. And that registered Phase 1 trial, begun in 2015 with forty-two healthy volunteers, was never updated after December 2015 and never posted results. The reporting indicates the team submitted data to the registry and then withdrew it before outside review.

None of that means BPC-157 does not work. It means the evidence base has not been built in a way that would let anyone find out.

The sceptical account tends to leave this out, and it is the strongest thing in the compound's favour.

A group at Chang Gung University in Taiwan, unconnected to Zagreb, has published a sustained programme on BPC-157 and tendon: outgrowth and migration of tendon fibroblasts in the Journal of Applied Physiology in 2011, growth hormone receptor upregulation in Molecules in 2014, VEGFR2 activation in the Journal of Molecular Medicine in 2017, and nitric-oxide-mediated vasomotor effects in Scientific Reports in 2020. A Chinese military medical group published the only full two-species ADME study. A Turkish group has shown vasorelaxation in human arterial tissue. And in 1995, scientists at Parke-Davis in Michigan independently replicated the colitis effect in rats and published it in the Journal of Pharmacology and Experimental Therapeutics.

A compound that is nothing but one lab's artefact does not usually survive a real pharmaceutical company running a full toxicology package and a randomised Phase 2 on it.

Thousands of people are taking this, and pretending otherwise helps nobody. But it is worth being precise about what that record can and cannot tell you.

Across nine subreddits there are around fifteen thousand posts about BPC-157 going back to 2014, with discussion peaking in 2025. What people say they use it for, in order: general injury recovery, then tendon, ligament and joint problems, elbow, shoulder, knee, Achilles, rotator cuff, which alone account for roughly twenty-eight per cent of all posts. Gut symptoms come a distant third at under ten per cent. This is, in practice, a musculoskeletal folk remedy.

In a decade, that community has produced exactly one poll with a usable result. In October 2020, 219 people answered whether BPC-157 had worked for them systemically: 145 said yes, 74 said no. Sixty-six per cent positive. That is the best number the self-experimentation world has ever generated on this compound and it deserves to be reported. It also deserves its caveats: self-selected, unblinded, no definition of worked, no dose or duration data. Three later attempts to do better all collapsed without publishing anything.

The more important finding is about the shape of the record. We counted threads by how they were framed: roughly 508 with an explicitly positive title against 59 explicitly reporting no effect, about nine to one. That alone could reflect a drug that works. But in July 2026 someone posted asking specifically whether BPC-157 had not worked for other people, an explicit call for non-responders. In that thread, positive testimonials outnumbered non-response reports twelve to two. The same pattern held in a January 2025 thread from someone whose shoulder bursitis had not improved: fourteen positive-only replies, one negative. When a non-responder asks whether anyone shares their experience, they get told it worked.

There is a second mechanism worth naming. Null results are routinely reattributed to the product rather than the drug: is my BPC bunk, did I get underdosed vials, wrong injection site, wrong protocol. Two per cent of all posts discuss counterfeit or underdosed product. This is an unfalsifiability loop. If it works, the peptide works, and if it does not, the vial was bad. Whatever else is true, the community record overstates the response rate by an unknown amount.

It also matters what BPC-157 is mostly used for. Tendon pain and gut symptoms are subjective, fluctuating and frequently self-limiting. People typically start a new protocol at their worst point, which is exactly when regression to the mean will flatter any intervention. Most users are simultaneously doing rehab. Placebo response in musculoskeletal pain is large and well documented. None of that makes the reports worthless. It means they generate hypotheses rather than confirm them.

Here the community record is genuinely ahead of the literature. The most frequently discussed adverse themes are not the ones you will find on any clinic page.

Fatigue and brain fog appear in about three per cent of posts. Injection site reactions in two and a half per cent, consistent with a note in the peer-reviewed literature that BPC-157 has been known to cause pain and necrosis when injected in an aqueous solution or physiological saline. Anxiety in 2.4 per cent. And a distinct cluster, going back to 2020, around anhedonia and emotional blunting, nearly two hundred posts and over a hundred threads with mood terms in the title. Heart palpitations appear in around a hundred posts, with more than twenty threads titled for them specifically. None of this appears in any published study.

Nor does it appear in the regulatory record, and that absence is itself instructive. The FDA's adverse event reporting system holds three reports for injectable BPC-157, searched to December 2025. Three, for a compound one analyst projects as a two-billion-dollar telehealth market. That is not evidence of safety. It is evidence that adverse events from unapproved, self-administered, grey-market products essentially do not get reported. Users obtained the product outside the medical system, often do not tell their doctor, and no clinician can readily attribute a symptom to a substance that is not on any list.

What the FDA's own toxicology review did find, in 28-day repeat-dose studies in rats and dogs, were changes in clotting time in both species and liver-associated signals. It noted that no carcinogenicity studies exist at all, and that immunogenicity for a fifteen-amino-acid peptide given by injection or nasally has never been formally investigated.

The most measurable hazard is not the molecule. It is the product.

In April 2026 researchers published an analysis of 6,441 consumer-market peptide samples from 203 vendors, including 416 of BPC-157. The good news first, because it contradicts the it-is-all-fake-powder line: only 3.5 per cent of BPC-157 samples failed to contain the stated peptide, median measured dose was 103 per cent of label, and median purity was 99.6 per cent.

The bad news is more serious. Across all peptides, between 42 and 71 per cent of samples failed basic quality benchmarks depending on the standard applied, and endotoxin was detectable in 15 per cent of the samples tested for it. Endotoxin is bacterial contamination, and people are injecting these preparations into their own tendons. The authors also found that purity did not predict endotoxin burden, so a clean certificate of analysis tells you nothing about whether the vial is safe to inject.

They were explicit that their dataset is biased favourably: samples are submitted voluntarily, so confident vendors submit and bad batches are under-represented. These are best-case numbers. Note also that this analysis is a preprint and has not yet been peer reviewed.

There is a separate and simpler problem. BPC-157 has been on the World Anti-Doping Agency's Prohibited List since 1 January 2022, under category S0, non-approved substances. Banned at all times, in and out of competition, with no therapeutic use exemption available.

We found three sanctions. A UFC fighter received a four-month suspension in 2023. She had self-reported, and had documentation that a doctor had prescribed it to treat a medical condition. It made no difference, because no exemption exists for a non-approved substance. A speedskater received a year in 2024. She never tested positive, but declared a supplement containing BPC-157 that a medical provider had recommended. A triathlete received four years in 2025.

Two of the three were self-declared rather than caught by testing, which means the sanction count is certainly an undercount of use. This applies to amateur and masters competitors in WADA-code sports, not only professionals, which covers a great many people in Irish athletics, rowing, cycling, swimming, triathlon and rugby.

Public figures get quoted constantly in this space, usually inaccurately.

Peter Attia devoted an August 2026 episode to peptides and placed BPC-157 in his unsupported category: three decades, dozens of fantastical benefits, and not a single human RCT, which he called the signature of a great marketing campaign based on hype and hope. He has no commercial interest in the answer.

Andrew Huberman is more open but is routinely misquoted as an endorser. His actual words include I am not suggesting anyone run out and take BPC-157, alongside an explicit warning that because the compound promotes blood vessel growth, someone with a tumour may be maintaining or accelerating its growth.

Bryan Johnson is the most misreported of all. He did experiment with BPC-157 and published a protocol for it, but it appears nowhere in his current published intervention list, and he has never said why he stopped. Affiliate sites still cite him as a current user.

Among those who do promote it enthusiastically, the ones we checked had commercial relationships with peptide suppliers, telehealth companies or supplement brands. That does not make them wrong, but it belongs in the frame.

Two things could change this picture, and they point in opposite directions.

The first is a real trial. In February 2026 a randomised, placebo-controlled, quadruple-masked Phase 2 began recruiting: 120 patients with MRI-confirmed grade II hamstring strains, subcutaneous dosing against a standardised rehabilitation background, with return-to-sport time and MRI injury volume as co-primary endpoints and an independent data safety monitoring committee. Primary completion is due February 2027. This is the first adequately controlled efficacy trial in BPC-157's history. The caveat is that its sponsor is a peptide manufacturer rather than a drug developer, and it is a single site.

The second is regulatory, and it is stranger. In July 2026 the FDA's Pharmacy Compounding Advisory Committee considered whether BPC-157 should be allowed for pharmacy compounding in the United States. The FDA's own scientists recommended against, citing insufficient evidence of effectiveness, poor characterisation of the substance, and immunogenicity risk. The committee voted 8 to 6 to recommend adding it anyway, a panel that had rarely if ever voted against staff. Reporting indicates the supporting votes came largely from members representing telehealth companies. The recommendation is non-binding and formal rulemaking would still be required. But if it succeeds, legal American supply would arrive without a single completed efficacy trial, which would remove what little commercial reason remained for anyone to run one.

None of this changes anything in Ireland. There is no European Medicines Agency opinion on BPC-157 and no HPRA safety communication about it. The Irish and EU regulatory record on this compound is simply empty. It remains an unauthorised medicinal product here.

Everything above is compatible with two competing explanations, and they are rarely stated side by side. It is worth being explicit about what we think and about what would prove us wrong.

The first explanation is that BPC-157 has a real but modest local tissue-repair effect, most plausible in tendon and soft tissue, working through the VEGFR2 and nitric-oxide angiogenesis pathway that independent laboratories have now replicated. On this reading the compound was never tested properly because it lost its commercial sponsor in a corporate sale in 2006 and has had no patent to fund a trial since 2018, and the people reporting benefit online are reporting something real that medicine has simply not got around to measuring. If this is right, we would expect the hamstring trial to show a difference on MRI injury volume, an objective endpoint that placebo cannot move.

The second is that the effect is mostly the ordinary machinery of recovery: rehabilitation, time, regression to the mean from the worst point, and a large placebo response in exactly the subjective, fluctuating complaints people use it for. On this reading the enthusiasm is sustained by a community record that suppresses null results roughly nine to one and blames failures on the vial rather than the drug, sitting on top of an animal literature that is three-quarters one laboratory using one dose. If this is right, the hamstring trial comes back null, or moves only the subjective endpoints while the MRI shows nothing.

Our own read is that the second explanation is more likely than the first, but not overwhelmingly, perhaps two to one. The single-lab concentration, the single dose level, the absence of any human pharmacokinetics and the undisclosed ownership interests are serious problems that would stop any pharmaceutical company at the first meeting. But the Parke-Davis replication in 1995, the independent Taiwanese mechanism work, and above all the fact that a real pharmaceutical company ran a full toxicology package and a randomised Phase 2 and got a positive trend, are not what a pure artefact usually looks like.

The useful thing about stating it this way is that both versions make different, checkable predictions, and one of them gets tested in February 2027. We have written down what we expect. When that trial reports we will say plainly whether we were right.

If BPC-157 turns out to work, we will have found out roughly two decades later than we should have, and not because anyone suppressed it. If it turns out not to, a lot of people will have injected an unregulated product into their tendons to find out. Both of those are worth saying, and we would rather say both than pick the one that reads better.