Are Peptides Good for You, or Safe? The Library Counted

"Are peptides good for you?" is usually asked about one vial, and answered about a whole class. Peptides are a chemical class, not a product: a peptide is any short chain of amino acids, which takes in insulin, the GLP-1 medicines, hormones the body makes every hour, and dozens of compounds sold online for research use. Asking whether peptides are good for you is a little like asking whether pills are. This page does not answer that for any reader, and it recommends nothing. It counts what this library holds, so the question can be asked about something specific.

The short answer, counted

On 3 October 2026 this library held 107 compound entries. Sorted by what each one is:

So there are at least three different questions hiding in one. For the approved medicines, "is it good for you" has a measured answer, for a named condition, with a measured list of risks. For the research-market compounds, the honest answer is that nobody has measured it in people. For the hormones, the question barely applies: the body makes them whether anyone buys them or not.

The groups are this site's reading of each entry's status field on that date, and two judgement calls sit in them: "approved in at least one country" includes GHRP-2, whose one approval is a Japanese diagnostic test, and the research-market group includes compounds the body also makes, such as glutathione and humanin, because what is sold is a manufactured vial. Which approval matches which marketed use is a separate question, set out on the library's index by purpose.

Where "good for you" has evidence: the approved medicines

The strongest answer to the question belongs to the 30 approved entries, and it comes with conditions attached. An approval attaches to an indication, a population and a dose, and it arrives with a label that states the risks as plainly as the benefit.

• Benefit, measured. Semaglutide and tirzepatide have controlled trials in thousands of people for blood sugar and weight; teriparatide for fractures; octreotide and lanreotide for hormone-secreting tumours. Each entry sets out the trials and the population.
• Risk, measured. Liraglutide's labels carry a boxed warning because it caused thyroid C-cell tumours in rats and mice, with human relevance not determined. Octreotide's label warns of heart-rhythm effects and gallstones. Teriparatide's lists osteosarcoma in rats as a warning and precaution.

That pairing is the point. An approved peptide is not "safe" in the abstract; it is a product whose benefits and harms have been counted, in a defined group of people, well enough for a regulator to weigh one against the other. The newest example is the clearest: elamipretide (SS-31) was approved on 19 September 2025 for muscle strength in Barth syndrome, an ultra-rare disease, under an accelerated approval built on 12 patients' safety data — and for nothing else.

Where it does not: the research-market best sellers

The compounds most people mean when they ask the question are the ones sold as research chemicals. Here is the same count for eleven of them, with two approved peptides and a common drug as controls. Searches were run on 3 October 2026; the method is at the end of this page.

† The registry returned two records for selank; both are brain-stimulation studies matched on an unrelated word, so the count shown is zero.

Three things in that table matter more than the totals.

"Tagged human" is not "given to humans." PubMed's human tag marks any paper involving people, including studies that only measured a substance already in the blood. MOTS-c has the largest human count of the eleven, 150, and its entry shows that almost all of it measures the body's own MOTS-c rather than giving it to anyone. The same caution applies to the randomized-trial tag, which marks the study design, not which compound was the treatment.

Not one of the eleven had a registered trial with results posted. The largest registered programme among them, ipamorelin's two company trials, studied recovery of bowel function after surgery, and the published trial found no significant difference from placebo. BPC-157's human record in PubMed, counted on its entry, is three small reports with 30 people between them.

The gap is wide. BPC-157, the most-sold of these compounds, has about one twenty-fifth of semaglutide's PubMed record and none of its trials. That is not a verdict that BPC-157 is harmful. It is a measurement of how little is known.

What FDA's staff and its advisers said in July 2026

Seven of these compounds went before FDA's Pharmacy Compounding Advisory Committee on 23 and 24 July 2026, as candidates for the 503A Bulks List — the list of substances US pharmacies may compound into medicines. FDA's own staff briefing documents, posted with the meeting record, reach the same conclusion for each: that "the evaluation criteria weigh against placing" the compound on that list.

The committee then voted against its own agency's staff on six of the seven. The tallies above are as reported by STAT (day one, day two), PharmExec and RAPS; FDA's meeting page carried no minutes when it was read on 3 October 2026. Three limits on reading the votes. A committee vote is advice: it does not put a substance on the list, and the list changes only if FDA completes rulemaking, a process trade press described as stalled in late September 2026. It is not an approval of any of the seven as a medicine. And it is not a finding that they work or are safe: the evidence the committee weighed is the evidence FDA's staff summarised — the BPC-157 briefing, for one, concludes there is "a lack of evidence to support the effectiveness" of the compound for ulcerative colitis, and notes that it is often unclear which form of the compound a published source actually tested.

What "safe" can and cannot mean here

For an approved peptide, "safe" means something checkable: a list of adverse effects, measured rates, warnings, and a manufactured product whose contents are controlled. For a research-market vial, three things are missing at once.

• No label. There is no document stating a tested dose, known interactions, who should not use it, or what to watch for.
• No defined product. A vial sold as BPC-157 or TB-500 is made to no pharmacopoeia standard; FDA's briefings repeatedly note uncertainty about which salt or form the literature used. What is in a given vial is a separate question, and one this library cannot answer.
• No human safety data to speak of. Where nothing has been reported, that is usually because nothing has been studied. Absence of reported harm is absence of measurement, not evidence of safety.

Reports of side effects people describe after using these compounds are gathered compound by compound on Peptifact's peptide side-effects page. Anyone weighing a specific compound for a specific reason is best served by a clinician, who can say what is known for that person — which is a question no count of papers answers.

How the counts were made

All searches were run on 3 October 2026. PubMed: title-and-abstract searches through NCBI's E-utilities for each name and its common spellings (for example BPC-157, BPC 157 and its development code PL 14736; for DSIP the full name and emideltide, without the bare acronym, which collides with unrelated terms), then the same search limited to the humans subject heading and to the randomized-controlled-trial publication type. Metformin is the control that shows the filters work. ClinicalTrials.gov: the count of records the API returns for each name in its intervention field; the selank and MOTS-c records were read one by one, and the entries for BPC-157, ipamorelin and GHK-Cu describe theirs. The library groups come from each entry's status field on the same date. The FDA conclusions are quoted from the seven briefing documents linked from the meeting page; the vote tallies are from STAT, PharmExec and RAPS, read on the same date.

Related entries: what peptides are, the full compound list, and the mitochondrial compounds sold for tiredness on peptides for energy.