BPC-157: Body Protection Compound

Made in the stomach, and it survives the stomach

The name is literal: Body Protection Compound. The 15-piece chain is a fragment of a larger protein that sits in human stomach juice, and its one genuinely odd physical property is that it does not fall apart in that juice for more than 24 hours. Almost no peptide survives a stomach. That single property is the only reason a swallowed form is discussed at all, and it points at the gut as the compound's home ground.

Nearly the whole research record traces to one group in Zagreb, Croatia, led by Predrag Sikirić, running since 1993 and now past 150 papers.

It is a coherent, decades-deep programme applying one consistent dosing scheme across dozens of tissues, and it is also most of the world's evidence coming from a single group with a stake in the answer. It is the reason independent trials in people matter more here than they would for a compound many labs had poked at.

What the different kinds of proof can and cannot tell you

Five studies in people, and the biggest one about an injury was a round of phone calls

FDA compiled the total published human experience for its July 2026 review. It is short enough to print in full.

The only randomised trial ever run on BPC-157 in a person was about an inflamed bowel, it exists only as a conference abstract, and the result range crosses zero — meaning the treated group and the untreated group cannot be told apart. And the only published report of BPC-157 doing anything for a musculoskeletal injury in a human is somebody reading old charts and then telephoning the people in them.

Eleven of twelve people who got BPC-157 alone in that knee series said they improved a lot. There was no dummy injection, no random assignment, no standard pain questionnaire and no imaging. That is exactly the design that cannot separate a real drug effect from the well-known fact that sticking a needle into a sore knee makes people feel better for a while.

Two registered trials matter. A 2015 safety and absorption study in 42 healthy volunteers in Mexico gave single swallowed doses of 1 mg, 3 mg or 6 mg, then 3 mg every eight hours for two weeks. It never posted results, and the results submission was formally cancelled in 2016. Nobody knows what it found.

The trial that could actually settle the tissue-repair question started recruiting in February 2026. It is randomised, double-blind, placebo-controlled, injected under the skin once a day for 14 days alongside a standard rehab programme, in 120 people aged 18–45 who have a grade II hamstring tear confirmed on a scan within 72 hours of doing it. The two things being measured are days until they can play sport again and the change in the size of the injury on the scan from day 0 to day 14. It is expected to finish in February 2027.

Until that reads out, there is nothing controlled in a human for any soft-tissue injury. Not weak data. None.

Rats: what happened, in which tissue, at what dose

The animals in every study below were young, healthy, and injured on purpose.

A rat Achilles torn off the bone came back mechanically stronger

The join between tendon and bone is the hardest soft-tissue repair in the body and the one that most often fails. In rats whose Achilles tendon was surgically cut off the heel bone, BPC-157 raised the load the repair could take before failing, its stiffness and its springiness, all significantly above untreated animals. Under the microscope the collagen fibres were better lined up, there was more type I collagen — the strong, rope-like kind — and more blood vessels had grown in. Dose was 10 µg/kg, 10 ng/kg or 10 pg/kg into the belly cavity once a day, starting 30 minutes after surgery.

The same paper carries the finding most relevant to anyone who has already had a cortisone shot: the steroid 6α-methylprednisolone consistently made the healing worse, and BPC-157 substantially undid that damage.

Cut a ligament, crush a muscle, or sever the join between them — all three healed faster

The ligament model cut through the medial collateral ligament in rats and followed them 90 days. Treated animals came out ahead on four separate measures at once: how the leg worked, how strong the ligament was, how it looked to the eye, and how it looked under the microscope. Three different ways of giving it each worked — injection into the belly cavity at 10 µg/kg or 10 ng/kg daily, a cream at 1.0 µg per gram rubbed on, and plain drinking water at 0.16 µg/mL.

The muscle model both cut through and crushed the rat calf muscle, and deliberately gave the animals a steroid to wreck the healing. The authors' finding is that BPC-157 completely reversed that steroid damage, given either by injection or as a cream, with the leg working normally again in treated animals.

The place where muscle turns into tendon — a classic site of grumbling strain that never settles — got its own study. In treated rats the injured leg's stiffening resolved completely, while untreated animals stayed disabled. The muscle wasting seen in untreated animals was prevented. By days 28 to 42 the new tissue ran in the right direction instead of being bridged by scar. Both the microgram and the nanogram doses worked.

The biggest recent study, from January 2025, pulled the quadriceps entirely off its bony anchors and then treated the rats only through their drinking water. The vastus muscles reattached as early as 21 days. Rectus femoris was back to normal anatomy at about 28 days. At three months the mature fibres ran parallel to the bone and sat right against it. Reattachment was tracked on ultrasound — the same imaging used on people.

The review that swept this whole literature screened 544 papers and kept 36. Thirty-five of the 36 were animal or lab work. Across muscle, tendon, ligament and bone the results point the same way: better function, better structure, better mechanical strength. The same authors put the counterweight in the same breath — harm is possible from unregulated manufacturing, contamination, and the fact that safety in people is unknown.

A cut sciatic nerve regrew faster, which is the nearest thing to a nerve finding

Rats with a completely severed sciatic nerve got 10 µg/kg or 10 ng/kg into the belly cavity, into the stomach, locally at the repair site, or loaded straight into the tube bridging a removed segment. Self-mutilation of the dead-feeling limb — the standard sign that a rat's nerve pain is bad — did not happen in treated animals. Electrical testing showed stronger muscle signals. Weekly walking tests improved. Counting fibres under the microscope found more of them, thicker, with better insulation around them, at one to two months.

It is a nerve cut clean through and sewn back together, not a nerve under slow mechanical compression. No study of BPC-157 in a compressed nerve has ever been published. The difference is not a technicality: a cut-and-repaired nerve regrows along a clear path, while a compressed nerve has to survive a mechanical problem that is still present. Different problems, different bottlenecks.

The two spinal studies squash the cord, and both used doses a hundred times higher

Two rat spinal-cord studies exist. Both compress the cord itself.

The 2019 study opened the spine at L2–L3 and compressed the tail end of the cord for 60 seconds, then gave a single injection into the belly cavity ten minutes later at 200 µg/kg or 2 µg/kg. Treated rats got progressively better tail movement, did not self-mutilate, and their spasticity had resolved by day 15, with less swelling and fewer dead motor nerve cells in the grey matter.

The 2022 follow-up is the more interesting one, because it waited before treating. After a one-minute squash by a 60–66 g weight that definitely paralysed the tail, one group got 2 µg/kg into the belly cavity at ten minutes. A second got a 10 ng/kg dose into the stomach on day 4. A third drank 10 µg/kg in water from day 4 to day 30. Recovery was described as fast and lasting out to a full year, with only patchy swelling, minimal bleeding and no loss of nerve insulation at day 30. Treatment started four days late still worked — which matters, because nobody turns up at a clinic ten minutes after their back goes.

The doses used in the spinal work run up to a hundred times the doses used in the tendon work, so no number carries across between them.

How the compound is supposed to build tissue back

BPC-157 raises the signals that grow new blood vessels at an injury. New vessels bring oxygen and repair cells to tissue that had neither. Alongside that, it pushes collagen production and gets tissue-building cells to crawl into the gap.

The receptor step has been worked out in detail. In the cells that line blood vessels, BPC-157 makes the cell pull VEGFR2 — the main docking point for the body's chief vessel-growing signal — inside itself, and then switches on the VEGFR2–Akt–eNOS chain over time. Block the pulling-inside step with a chemical called dynasore and the whole effect disappears, which means the pulling-inside is required, not incidental. Vessel density went up both in a dish and in living animals, and blood flow came back faster in muscle that had been starved of it.

The crawling step has its own named mechanism. In tissue-building cells taken from rat tendon, BPC-157 dose-dependently added phosphate groups to two proteins, FAK and paxillin, which are the machinery a cell uses to grip a surface and haul itself along. The total amount of those two proteins did not change; only the switched-on fraction rose. The same study found the cells were not making more of themselves, so the effect is cells moving and surviving, not cells multiplying in number.

A third mechanism sits underneath both. Tendon cells given BPC-157 at 0.5 µg/mL grew about seven times more growth hormone receptors by day three, measured both as the instruction and as the finished protein. The cell does not get more growth hormone. It gets better at hearing the growth hormone already going past.

Nitric oxide — the gas that makes vessels widen — is pushed up in some situations and down in others, always paired with mopping up damaging free radicals. The Zagreb group reads that two-way behaviour as the whole point of the molecule: it pushes tissue toward normal rather than in one fixed direction. FDA reads the same observation as the reason no dose-response curve has ever been drawn for it. Both readings fit the data.

Read the chain against the tissues it has been tested in. New vessels: shown in cells and in rats. Vessels feeding a starved tissue: shown in muscle. Cells restarting production of the molecules that hold water in a matrix: never studied for this compound, in any tissue.

The painkiller you are probably already taking runs the opposite way to expected

Most people who arrive with back pain are already on an anti-inflammatory, and most compounds interact badly with those. This one was studied as an antidote to them.

Diclofenac at 12.5 mg/kg into the belly cavity once daily for three days produced severe stomach, gut and liver damage in rats, raised bilirubin, AST and ALT, made the liver heavier, and caused prolonged drowsiness that progressed to liver-driven brain failure with brain swelling and damaged nerve cells. BPC-157 blocked that entire cascade at both the microgram and the nanogram dose, given either by injection or in drinking water.

Read carefully, that is a gut-and-liver protection finding, not proof that combining the two heals a tendon better. What it does mean is that the medicine most likely to already be in the room is not, in rats, a documented problem.

The steroid interaction is the more consequential one. Two separate studies show BPC-157 undoing the damage a corticosteroid does to healing tendon and muscle. In animals a steroid injection measurably impaired the repair it was given for, and this compound reversed that impairment.

It is gone from your blood in about fifteen minutes

The only published measurements of what the body does to this compound come from rats and beagles.

Average time to clear half of it: 15.2 minutes. In dogs given it into a vein, 5.27 minutes. Peak level in the blood arrived three minutes after an injection into muscle. How much reaches the blood after an injection into muscle (bioavailability) was 14–19% in rats and 45–51% in dogs — wildly different between two species, and far from all of it. It leaves in urine and bile. Levels in kidney, liver, stomach wall, thymus and spleen went well above the level in blood.

A compound cleared that fast cannot be working by keeping a steady level in your blood. Whatever it does, it does fast and locally, then sets off a slower tissue process that outlasts the exposure by weeks. Two practical things follow. First, that is the argument for splitting a daily dose rather than taking it all at once. Second, a once-a-week schedule has nothing behind it.

No measurement of any of this has ever been made in a person. Every number above comes from a rat or a dog and is carried across species by assumption.

Turning powder into a dose

BPC-157 arrives as a freeze-dried white powder in a sealed glass vial, usually 5 mg or 10 mg. You add liquid to it. The liquid is bacteriostatic water — sterile water with 0.9% benzyl alcohol in it as a preservative, which is the thing that lets you put a needle into the same vial repeatedly over weeks. Plain sterile water dissolves the powder fine but has no preservative, so that vial is a one-time proposition.

The measuring device is a U-100 insulin syringe. U-100 means 100 units in a millilitre, so one unit on the barrel is 0.01 mL. Every conversion you will ever need comes from two lines:

• Micrograms per mL = total micrograms in the vial ÷ mL of water you added.
• Micrograms per syringe unit = that number ÷ 100.

Worked through: a 10 mg vial holds 10,000 mcg. You add 2 mL of bacteriostatic water. 10,000 ÷ 2 = 5,000 mcg per mL. 5,000 ÷ 100 = 50 mcg in one unit. A 250 mcg dose is therefore 250 ÷ 50 = 5 units on the barrel.

Adding more water does not add or remove peptide. The vial holds what it holds. More water only changes how many syringe units carry a given dose. More water gives you bigger, easier-to-read unit counts and a slightly bigger squirt going in. Less water gives you a smaller sting and finer lines to misread. Below about 5 units per dose, the error in reading a U-100 barrel starts to swamp the dose itself — which is the practical argument for putting 3 mL rather than 2 mL into a 10 mg vial.

Ruining the peptide by accident is easy. Let the vial come to room temperature first. Run the water slowly down the inside wall of the glass rather than blasting it onto the powder. Swirl until it dissolves. Never shake it — shaking tears at the surface where liquid meets air, which unfolds peptide chains and makes them clump together, and clumping is the exact thing FDA named as the trigger for the immune system reacting to it (immunogenicity).

A mixed vial lasts about four weeks

Do the arithmetic against your own plan before you buy. A 10 mg vial at 250 mcg a day is 40 doses. Four weeks is 28 doses. You will throw away roughly a third of the vial, or you must buy a smaller one, or take more per day. There is no fourth option — refreezing it is the one thing that definitely damages it.

No study compares injecting near the injury against injecting anywhere else

Two placements compete and the animal work cannot referee between them, because rats were dosed into the belly cavity, into the stomach or rubbed on the skin, almost never at a specific human-style injection site.

• Under the skin, near the sore part. What most people do. The argument is a higher concentration where the repair is needed, which fits a 15-minute clearance and fits tissue levels running above blood levels.
• Under the skin of the belly, anywhere. Standard for gut, nerve and organ targets, and easier to do to yourself. The argument is that a compound reaching kidney, liver and stomach wall above blood level obviously travels, so the entry point may not matter.

The evidence tilts slightly toward the entry point not mattering. Several of the strongest structural results in the entire literature — the cut ligament, the muscle-tendon junction, the quadriceps reattachment, the sciatic nerve, the spinal cord — were produced in rats that simply drank the compound and still repaired a specific injury a long way from their gut. Nothing anywhere in the record shows a local injection beating a systemic one head to head.

Here are the doses that exist, all of them either from rats or scaled up from rats:

There is no validated human dose, no validated cycle length, no taper and no ceiling. USADA states it without softening: nobody knows whether a safe dose exists, or whether there is any way to use this compound safely for a named condition.

Every dose ever given to a human in a published study is measured in milligrams. Scaling the standard rat dose up to an adult lands at roughly 0.7 mg. The regimens circulating on forums and printed on vendor sites are measured in micrograms — a thousandth of a milligram — and land around 250 to 500 mcg. That is roughly a hundredth of anything ever formally given to a person. Nothing in the literature supports the microgram regimens. Nothing in the literature refutes them either. The dose question is not contested; it is simply open, and everyone taking it is guessing.

The tablet and the vial are not the same product

Swallowed BPC-157 exists as capsules or tablets, commonly 500 mcg each, and the stomach-survival finding is what makes that format arguable at all. But surviving your stomach acid and getting into your bloodstream are two separate questions, and only the first has ever been shown in a person.

The evidence is friendlier to the swallowed route than the injection convention implies. Several of the strongest structural repair results in the whole corpus — ligament, muscle-tendon junction, muscle reattached to bone, delayed spinal cord recovery — came from rats drinking it in water. What nobody has is a measurement of how much of a swallowed dose gets into a human's blood, so choosing between the two forms rests on rat route data and on convenience, not on anything measured in a person.

How long anything would take, if it worked

Timelines here are set by tissue biology, not by the compound. Collagen remodels at the speed collagen remodels. Nerves regrow at about a millimetre a day. Nothing changes that.

No human timeline has been established, because no controlled human trial has finished.

The honest inference: these are collagen and nerve timescales, four to twelve weeks. A result reported inside a fortnight is pain relief, not repair. Pain relief arrives before tissue has been rebuilt, so a change in symptoms at day 10 says nothing about the structure.

FDA's own toxicology is more specific than the safety talk around it

The Zagreb group's stated position is that BPC-157 is very safe, with no side effects in trials and a lethal dose never reached in toxicology work. FDA read the same body of work and described it differently.

In 28-day repeat-dose studies injecting into muscle, rats showed shortened clotting time and dogs showed lengthened clotting time on the same test — an unexplained effect on clotting that runs in opposite directions in two species. Blood ALT, glucose and triglycerides went up. No cancer-causing study of BPC-157 in either of its two chemical forms has ever been done. It did not damage DNA, and no birth-defect signal appeared when it was injected through gestation days 6–15.

FDA raised four objections. They are independent of each other, which means answering one leaves the other three standing.

The cancer question has a real answer and a real hole in it

The worry is mechanically sensible. A tumour cannot grow past a millimetre or two without recruiting new blood vessels. A compound whose headline action is growing new blood vessels could, in principle, supply them. An independent 2025 review states the worry and lands on the practical version of it: a cancer nobody has found yet.

The Zagreb group published a formal rebuttal in the same journal. Their case is that BPC-157 strongly blocked vessel growth into the cornea, that it has shown anti-tumour activity both in living animals and in a dish, and that its nitric oxide effect is always paired with mopping up free radicals — that it pushes vessel growth toward the normal, self-limiting kind rather than driving it indiscriminately.

Where that leaves you: no published rodent study shows BPC-157 speeding up tumour growth. No cancer-causing study exists. There is zero human cancer safety data. An active or suspected cancer is the one situation the mechanism itself argues against, and the missing study means the argument cannot be closed either way. If you have had a cancer, this is the question to take to your oncologist rather than to a forum.

FDA said no, and its own advisory committee said yes by two votes

The regulatory position moved twice during 2026, and most summaries you will find online are stale.

If a page you are reading still says BPC-157 is FDA Category 2, it has not been updated since 14 May 2026.

One detail in the meeting agenda changes what that vote means. The only condition FDA evaluated was ulcerative colitis. Tendonitis was among the uses put forward, and FDA declined to evaluate it because there was not enough information. A favourable vote about a bowel condition is not a favourable vote about tendon or disc repair, and it is being quoted as though it were.

The committee itself was not persuaded of much. One panelist who voted yes described the thing being voted on as a black box. One who voted no pointed out that roughly 30% of people given a dummy treatment report improvement — sometimes even on objective measures — which is the entire difficulty with the uncontrolled reports the nomination rested on.

Reporting on the meeting noted that a majority of the panelists voting yes had ties to the peptide industry.

The vote creates no authority. It is advisory, FDA is not bound by it, and a formal rulemaking cycle realistically runs eight to twelve months.

There are two legal ways to compound a peptide and BPC-157 is on neither list

503A covers traditional compounding pharmacies. 503B covers registered outsourcing facilities, and it is a separate list with a separate standard: an outsourcing facility may not compound with a bulk substance unless that substance is on the 503B list, or the finished medicine is on FDA's drug shortage list at the time. BPC-157 is not on the 503B list, and the July 2026 vote concerned 503A only.

That same page sets out what FDA expects of any lawfully compounded substance: a valid certificate of analysis, manufacture by an establishment registered under section 510 of the FD&C Act, and compliance with any applicable USP or National Formulary monograph. Grey-market BPC-157 meets none of the three.

If you compete in anything, this is a strict-liability problem

BPC-157 is named explicitly on the WADA 2026 Prohibited List under S0, Non-Approved Substances, banned at all times — not just on competition day. S0 covers any substance with no current approval by any government health authority for human use anywhere, which is exactly what BPC-157 is.

Strict liability means you are responsible for what is in your body regardless of intent and regardless of who recommended it. That applies at college, professional and masters level, in sports most people do not think of as drug-tested.

The Department of Defense goes further. Its Operation Supplement Safety programme lists BPC-157 on the Prohibited Dietary Supplement Ingredients List and states the position flatly: BPC-157 is not a dietary ingredient, it is an unapproved drug. That applies to service members however it is taken.

"Research use only" is a shipping label, not a legal category

Nearly all BPC-157 sold to the public is labelled a research chemical, not for human consumption. That phrase creates no lawful route to a person. It is a disclaimer attached to a product with no legal standing as a medicine, a food or a supplement in the United States. USADA names the tell directly: sites that print "research only" while also printing human dosing instructions are describing what they expect to happen.

The purity problem is the concrete risk, and it is bigger than the legal one. Independent testing of grey-market peptide samples finds mislabelling, wrong dosing or contamination in roughly 30%, and the contaminant classes are specific: bacterial toxins, incomplete chains that are missing pieces, leftover solvents, heavy metals, live microbes and breakdown products. Incomplete chains and clumping are the same two failure modes FDA named as the things that make an immune reaction more likely — which means the supply-chain risk and the drug risk are the same risk wearing two hats.

30% contained incorrect amino acid sequences and two-thirds of samples fell below the 95% purity threshold.

A certificate reading "99% by HPLC" does not clear a batch, and understanding why is worth two minutes. HPLC confirms the target peptide is there and measures how much of the total it is. It says nothing about what the other 1% consists of. An incomplete chain or a fever-causing bacterial toxin sits inside that remainder and never shows up. What you want is HPLC and mass spectrometry, done per batch, by a laboratory that is not the seller. A vendor's own certificate is not third-party testing, and a certificate photographed from a different batch is not a certificate at all.

Which problems the evidence actually reaches

Every positive animal model in this literature is an acute surgical injury — a tendon cut through, a ligament severed, a muscle crushed, a tendon torn off bone, a nerve cut. There is no rat model of a fifteen-year grumbling tendon, and no reason to assume a repair signal helps tissue whose problem is failed remodelling rather than an unfilled hole.

The cortisone-reversal row is the most useful finding in the entire animal literature, because a large share of people with a bad back arrive having already had a steroid injection that measurably worsened the healing they came in for. If any part of this compound's animal record applies to you, that is the part with the clearest logic behind it.

The long-standing-tendon-pain row is weak exactly where demand is highest.

Fifty people said what happened to them, and here is the count

This page holds 50 first-person accounts posted publicly between 2015 and 2026 — 40 on Reddit across r/Peptides, r/steroids, r/backpain, r/Sciatica, r/climbharder, r/PEDs, r/Biohackers, r/moreplatesmoredates, r/PeptideGuide, r/Peptidesource and r/Supplements, and 10 on X. Every one is self-reported by a person you cannot identify or question. Nobody assayed their vial, so the substance in the syringe is unknown in every single case. Nobody was randomly assigned. Almost nobody scanned the injury before and after. People who got better write posts more often than people who did not, so the set is filtered before you even read it.

It is here for one reason. No controlled trial in a human musculoskeletal injury has ever finished, so this is the entire record of people using it for the reasons you are considering it — and it is what you have already read by the time you got to this page.

The count, with the denominator

Of the 50 accounts, 13 report no personal result at all — they are protocols, questions, general threads about the science, or one person's dosing accident. That leaves 37 accounts that state what happened to the person writing. Here is the split:

Half of the people who reported an outcome said it helped, and half did not. If you have seen this compound described online, you have almost certainly seen a version of that record where the 14 accounts reporting nothing or harm were left out. That version is a lie by selection, and it is the standard version.

Four more things the raw count hides.

The 10 "nothing happened" accounts are not shorter courses or smaller doses than the successes. Several ran six weeks or more at the same 500 mcg a day that the successes ran.

The 4 harm reports describe things the rat toxicology gives no reason to expect and the five human studies were far too small to catch: a joint that got worse and inflamed, vision going blurry over weeks, a finger swelling with fluid, and leg pain returning during treatment for a herniated disc.

One account reports a ten-fold dosing accident, which is a safety datapoint of its own and the reason the reconstitution arithmetic earlier on this page matters more than it looks.

And exactly one account in 50 reports an objective check by a clinician. One. Everybody else is reporting pain, movement and confidence — the three things most responsive to time, rest and expectation.

The 18 who said it helped

The account crediting a clinician with saying the tear was gone is the single best entry in the whole set, and even that was a re-examination, not a repeat scan. And the account that asks itself "is this placebo?" also added collagen and more protein in the same window, which the writer says outright — that honesty makes it more useful than the confident ones, not less.

The 5 who got part of a result, or lost it

The last one is a pattern with no counterpart anywhere in the published literature: the effect fading on a steady dose, then not returning when the dose was raised. Nothing in the rat work predicts that. Nothing in the pharmacology explains it.

The 10 who said nothing happened

These are the entries marketing copy leaves out. They are the same doses, the same durations, and several of them are longer courses than the successes.

The standard reply offered inside these threads to a negative report is that the person got a bad vial. That reply cannot be checked — nobody in this set tested their product — and it is the exact mechanism by which a grey-market compound builds a record that can never fail. Every success counts as evidence; every failure counts as a counterfeit.

The 4 who got worse

The first is a harm report during a course taken for a herniated disc: leg pain returning and worsening on treatment.

The second describes one joint improving while another got inflamed and worse in the same person, on the same dose, in the same course.

The dosing accident, and where the numbers came from

That is ten times the intended dose, caused by misreading a syringe barrel. Nobody was hurt. It is on this page because the numbers everyone uses have nothing under them, and the people using them say so plainly:

Almost every account in the set describes 250 to 500 micrograms a day, under the skin, sometimes split into two or three shots, run for four to eight weeks. That band converged by copying, not by measurement. It is about a hundredth of the milligram doses used in every published human study.

The 13 that report no outcome at all — including the six loudest posts here

Thirteen of the 50 accounts state no personal result. Seven are the protocols and dosing arguments already shown above. The other six are X posts, and they are worth looking at separately, because they are almost certainly the format in which most people first meet this compound.

Four are summaries of the animal literature written in the voice of personal endorsement — "repaired 5 different tissues", "no compound heals as many tissue types" — with a dose attached at the end and no report of the writer's own outcome anywhere in them. One is a person asking for an alternative. One is a person saying they never took it but wish they had.

They are in this count because leaving them out would be dishonest, and they are separated out because counting them as successes would be worse. The most confident writing about this compound contains the least first-hand information in it, and that is the single most useful thing this section can tell you.

Where the accounts contradict the animal work

The animal record is strongest for fresh injuries and weakest for long-standing tendon pain.

The forum record inverts that. Long-standing elbow, wrist and shoulder tendon pain of one to five years' standing is the most common thing people treat, and spinal complaints — where no animal model exists — generate some of the most emphatic success reports.

The most emphatic of those accounts describes pain easing over six to seven weeks of injections after everything else had failed, and it is the one with the least behind it, because no animal study covers that tissue.

Two readings fit and the accounts cannot separate them. Either the compound reaches structures the animal work never tested, or long-standing pain that comes and goes was going to improve anyway and the compound arrived during a good month. Spontaneous improvement over roughly six to twelve weeks is well documented for several of these complaints, and that is exactly the window these protocols run.

The swallowed route contradicts the pharmacology the same way. Surviving stomach acid is proven; getting into a human's blood from the gut is not. People report the two forms doing different jobs, which is precisely what you would expect if the tablet works locally in the gut and never reaches a tendon at all.

The stacking problem, which is the biggest hole in the whole set

Almost nobody takes this alone. TB-500 is combined with it so routinely that a large share of these accounts cannot attribute any outcome to either compound. Others add growth hormone, testosterone, surgery, physiotherapy, diet changes and rest inside the same weeks.

The three most confident recovery stories on this page — the rotator cuff tear a clinician later called gone, the wrist rebuilt over seven vials, and the shoulder that felt 75% better in five days — were all run as BPC-157 plus TB-500, and two of them alongside structured rehab. Not one of them can tell you which part did the work.

It means a good share of the 18 positive accounts are reports about a combination, run alongside the two interventions with the best evidence in all of musculoskeletal medicine — load management and time.

What the counted record settles, and what it does not

It does not settle whether it works. Thirty-seven uncontrolled accounts filtered toward success are not weak evidence for a treatment effect; they are a different kind of object, and collecting three hundred more would change nothing about that. A 49% success rate in this set is not a 49% chance it works for you. It is a measurement of what people wrote down.

It does settle three things the published literature does not contain at all, and these are the reason it is worth counting rather than dismissing.

First, the real exposure. People are running 250–500 mcg a day under the skin for four to eight weeks, usually stacked with TB-500. No trial has ever studied that exposure. When somebody eventually does, this is the regimen they should test, and nobody would know it from the papers.

Second, harms at that exposure. Four of 37 accounts report getting worse — local swelling, a joint becoming more inflamed, blurred vision, and returning nerve pain. None of those appear in any published human safety dataset, because those datasets total about 130 people across five studies and would not detect a 1-in-10 problem if it existed.

Third, the mismatch. The two things people most often treat with it are long-standing tendon pain, which the animal literature supports least, and spinal complaints, which it does not cover at all.

The Phase 2 hamstring trial reading out in 2027 will answer the works-or-not question for one fresh injury in one muscle. Nothing currently running covers any spinal tissue.

What is settled about BPC-157, and what is not

The verdict, stated so it decides something. For a fresh tear or detachment in tendon, ligament or muscle — and most of all for repair that a corticosteroid injection has already impaired — BPC-157 has the strongest animal case of any compound written up here: consistent across at least six injury models, mechanically explained down to the receptor step, and reproduced by three separate ways of giving it. Taking it for that, at 250–500 mcg a day under the skin for four to eight weeks, is a bet on rat evidence that is worth naming as a bet: no controlled human trial in any of those tissues has finished, and the first one reads out in February 2027. For long-standing overuse tendon pain, and for anything to do with a spinal disc, the answer is no — not "unproven", no. Every positive study in the whole literature is a fresh surgical injury, there is no animal model of a grumbling tendon, the two spinal studies squash a cord at up to a hundred times the tendon dose, and the disc has never been studied in any species. Those two are what it is most often bought for, and they are the two the evidence does not reach.

BPC-157 is not an approved medicine in any country. Everything sold to the public is grey-market and unverified. Nothing here is a dosing or treatment recommendation.