Verified July 2026 · Cited to primary sources

Are peptides safe?

It depends on the peptide, and honest human safety data is thin. Only 14 of the 46peptides we grade have well-characterized human safety. For most people the real danger isn't the molecule; it's the unregulated gray-market supply the FDA has flagged for dosing errors and impurities.

Are peptides safe to take?

There is no single answer, because “peptides” covers everything from an FDA-approved drug with decades of data to a vial of powder that has never been tested in a human. Two questions decide the real risk: does this specific peptide have human safety evidence? and can you get a version whose dose and purity are actually verified? For most peptides sold today, the honest answer to both is no.

Which peptides have real human safety data?

We assign each peptide a safety flag (green, amber, or red) alongside its evidence grade. Green means well-characterized human safety: an FDA-approved drug or a long clinical history. Amber means limited human safety data with no major documented harms. Red means documented serious harms. Of our 46 graded peptides:

The uncomfortable pattern: most peptides sit at amber not because they are proven safe, but because there is too little human data to know either way. Absence of documented harm is not the same as evidence of safety.

Why is the gray-market peptide supply the real danger?

Even if a peptide were perfectly safe in theory, most people are buying it from a source that cannot prove what is in the vial. The FDA has documented dosing errors, impurities, and unapproved salt forms in compounded injectables, with adverse events serious enough to require hospitalization. A historical FDA survey of compounded drugs found that 31% failed standard potency testing. And independent testing of “research only” peptides routinely finds purity far below the 98%+ that regulated pharmacies target.

This is the vocabulary the community already uses: underdosed vials, bunk product, bait-and-switch, fake certificates of analysis. A COA from the seller proves nothing; the chromatogram can be recycled or generated. That supply-quality problem is exactly why we grade legal access routes and refuse to link gray-market vendors.

Does the “research use only” label mean anyone has checked the vial?

No, and that is the point of the phrase. A research-use-only label is not a quality claim, not a sterility claim, and not a statement that any regulator has looked at the product. It is a disclaimer about intended use, and the FDA has now said on the record that it does not even accomplish that.

On April 7, 2026 the FDA posted warning letters from its Center for Drug Evaluation and Research, dated March 31, 2026, to peptide websites. Writing to Gram Peptides (MARCS-CMS 721806), the agency said that “despite statements on your product labeling marketing your products for ‘Research Use Only,’” the website's own description of appetite suppression and weight loss established that the products were “intended to be drugs for human use.” It reached the same conclusion for Prime Sciences (721805) and Mile High Compounds (721600), for Pink Pony Peptides (721088) and for PekCura Labs (721709). Five sellers, five different disclaimer wordings, no different effect.

The safety-relevant detail is what the FDA said about the diluent. Because these sellers offered bacteriostatic water next to peptides that need reconstituting, the agency held that selling them together “demonstrates that you intend for your ‘Bacteriostatic Water for Injection’ to be used in combination for injection,” making the water a drug too. Read plainly, that is the FDA describing the ordinary research-peptide purchase, powder plus diluent, as a product intended for injection into a person, while noting that injectables “bypass some of the body's key defenses against toxins and microorganisms.” Nothing in that transaction involves a manufacturing standard, a sterility test, or a regulator who has seen the vial.

One seller has now been sentenced rather than warned, and the count he pleaded guilty to is a safety fact rather than a paperwork one. The United States Attorney's Office, Northern District of Indiana announced that the owner of Paradigm Peptides was sentenced to 70 months in federal prison, with a money judgment of $5,000,000, after pleading guilty to introducing unapproved new drugs into interstate commerce with the intent to defraud and mislead. Prosecutors said the business told customers the business was licensed and registered with the FDA; the business manufactured its products at its own laboratories in the United States; the products were safe, pure and tested; the products were approved by the FDA, while importing the goods from China, India and other countries. United States Attorney Adam Mildred put the harm this way: “These customers trusted them and consumed the products in the hope of improving their health or physique and instead were poisoned with known and unknown substances that were illegally imported from China, India, and other countries.” An estimated 54,000 people bought from that website believing the opposite of each of those claims. Full detail is on our are peptides legal page. For safety purposes the lesson is narrow and hard: a certificate of analysis, a claimed US laboratory and a claimed FDA registration are all things a seller can simply assert, and in this case a federal court found that is exactly what happened.

A third kind of record opened on August 12, 2026, and it is worth reading for one narrow safety reason. Eli Lilly and Company filed 6 federal lawsuits against US sellers of retatrutide, an investigational compound no regulator anywhere has approved, and Legendary Peptides, LLC, Texas Peptides Inc. and Lone Star Peptide Co. LLC trade under explicitly peptide-facing names. The safety point is not the litigation. It is that a product being sold, shipped and injected today can be one whose maker has never submitted it to any regulator for approval, and that the seller and the manufacturer are frequently not the same party. These are unproven allegations and no court has ruled. The docket detail, the causes of action and the source are on our are peptides legal page.

What did independent testing find inside the vials?

One figure has carried most of the July 2026 coverage: that almost 30% of tested peptide vials came back mislabeled, misdosed or contaminated. It is worth knowing where that number comes from, because it is repeatedly cited without its source and without its limits, and both change what it can tell you.

The figure is a quote. Rina Dukor, co-founder of BioTools, a lab that has tested pharmaceutical products for drug companies and federal agencies, told NBC4 Washington on July 23, 2026 that almost 30% of the peptide samples her team had analyzed were mislabeled, underdosed, overdosed, or contaminated with toxins or bacteria. Consumers send their vials to a testing company, Finnrick, and get back what is actually inside. In the same interview she added: “Among the thousands of tests we are doing, we do find either live or dead bacteria in a lot of the vials. That could lead to hospitalizations.”

Finnrick publishes its own cumulative count. Read on August 5, 2026, its site reported 9,537 vials tested, stamped current as of August 2, 2026, with public per-compound boards:

• Retatrutide: 3,495 tests across 239 vendors, board updated July 22, 2026.
• Tirzepatide: 2,334 tests across 183 vendors, board updated July 22, 2026.
• BPC-157: 712 tests across 103 vendors, board updated July 22, 2026.
• Semaglutide: 338 tests across 57 vendors, board updated July 13, 2026.
• CJC-1295: 288 tests across 54 vendors, board updated July 13, 2026.
• Melanotan II: 170 tests across 40 vendors, board updated July 10, 2026.

Those per-compound counts are the compounds the lab ranks publicly. They do not add up to the cumulative total and must not be summed, for the same reason the recall counts below cannot be: one tested vial can appear on more than one board.

Here is the part almost every retelling drops. The almost 30% was said on July 23, 2026; the 9,537is the lab's count as of August 2, 2026. The broadcast never stated the denominator behind the percentage, so the two numbers are ten days apart and cannot be multiplied together. You will see “30% of thousands of vials” written as though someone measured it. Nobody published that calculation, and we are not going to invent it here.

Sources: NBC4 Washington, July 23, 2026 for the figure and the quote, and the lab's own published boards for the counts. Neither is a peer-reviewed study, and we have found none on this question. If one is published, this section changes.

How many peptide products has the FDA actually recalled?

More than the headlines suggest, and from a source most people do not expect. We queried the FDA's own drug enforcement (recall) database on 2026-07-31, one compound at a time, matching on the product_description field. Every number below is a live count you can re-run; the database reported itself last updated on 2026-07-22. Searching the literal word “peptide” returns only 2, which is why the recall record is usually described as almost empty. Searching the compound names tells a different story.

These counts overlap and must not be added up. A single recall of a combination product, for example “Sermorelin Acetate 6mg / Ipamorelin 6mg,” is counted under both names. We report them as measured, per compound.

Two things in that table matter more than the totals. The first is who is being recalled. The peptide with the longest recall history by far is sermorelin at 92 entries, and sermorelin is the one peptide on this site with a routine legal route. Those recalls are not gray-market vials. They are licensed compounding pharmacies, and the reason attached to almost every one of them is the same phrase: lack of assurance of sterility. Nearly all were voluntary, firm-initiated, and classified Class II, meaning the FDA judged that exposure could cause temporary or medically reversible harm. Most were caught by inspection rather than by someone getting hurt. The legal route is the safer route, and it still fails often enough to fill a database.

The second is the empty rows. TB-500, Semax and Epitalon return no recalls at all. Read that carefully, because it is the easiest number on this page to misuse. It is not a clean safety record. A recall is an action a regulator takes against a manufacturer it regulates, so a compound that no licensed pharmacy is lawfully permitted to make cannot generate one, however much of it is being injected. The blank line measures the absence of a regulated supply chain, not the absence of harm. If you see a vendor cite “zero FDA recalls” as evidence that a peptide is safe, that is what the number actually means.

The most recent peptide recalls in the database are both still open, and both from the same firm:

• D-0051-2026, GenoGenix LLC (Boca Raton, FL): GHK-Cu (Copper Peptide) for Injection, all strengths and presentations. Class II, lack of assurance of sterility. Initiated 2025-07-30, classified 2025-10-21, status Ongoing, 221 units.
• D-0076-2026, GenoGenix LLC (Boca Raton, FL): Low Solubility Peptide Reconstitution Solution for Injection, all strengths and presentations, also labeled as manufactured for CELIA Health. Class II, lack of assurance of sterility. Initiated 2025-07-30, classified 2025-10-21, status Ongoing.

Neither was triggered by a reported injury. Both are sterility assurance failures, which is the failure mode this whole record keeps returning to: not a peptide that turned out to be toxic, but a sterile injectable that nobody could prove was sterile. Source: openFDA drug enforcement database. 9 of the 12 compounds we searched have at least one recall on record.

The recall the database does not have yet

Everything above comes out of the enforcement database, and there is a limit to that method we have to state plainly, because we ran into it. On 2026-08-05, Victory Medical Center Pharmacy of Austin, TX announced a recall of Compounded Glutathione 200 mg/mL Multi-Dose Vials sold under the brand name VMC, because pharmacy testing identified elevated bacterial endotoxin levels. The recall is voluntary, nationwide, to the patient level. On 2026-08-12 we queried the same openFDA endpoint every number on this page comes from for that firm. It returned HTTP 404, NOT_FOUND, no matches, with the database stamping itself last updated 2026-08-05.

The recall is real and the database is silent, and both are true at once. An enforcement number is assigned when the FDA classifies a recall, which happens weeks after the firm announces it. So a database query run in the window between those two events returns exactly what it would return if nothing had happened at all. That is the second way a zero on this page can mislead you, and unlike the first one it can catch us as easily as it catches a vendor. When we say a compound has no recalls, we mean no classified enforcement record as of the date stamped on the query, and nothing more.

The affected product is three lots, 1980571, 1981940 and 1984345, all expiring 2026-08-27, distributed to consumers in Texas, Florida and New York. The firm says it has already received reports of adverse events including fever, chills, severe headache, nausea, vomiting, tachycardia, changes in blood pressure, body aches, and instances of injection site reactions. Its risk statement is that there is reasonable probability that injectable products with elevated endotoxin levels could cause significant medical events including fevers, hypotension, inflammatory reactions, anaphylactic shock and death. Source: the FDA posting of the company announcement.

Why a glutathione recall is on a peptide page at all

Because glutathione is a peptide. It is a tripeptide, three amino acids, and when a newspaper headlines this event as a pharmacy recalling injectable peptides it is being literally accurate. That is worth sitting with, because it is the direction this whole category is moving. The compounds we grade on this site are the ones with a research literature and a docket number. The injectables actually going into people at scale, in med spas and wellness clinics, are a different and much larger set, and glutathione is near the top of it. As the word peptide widens to cover both, the safety record of the second set starts getting read as the safety record of the first.

We are not going to resolve that by narrowing the definition to suit us. Glutathione now has its own row in the table above, at 77 recalls, read 2026-08-12. That is more than any compound we grade except sermorelin. If you are deciding whether an injectable peptide from a compounding pharmacy is safe, that number belongs in the answer.

There is also a precedent, and it is close enough to be uncomfortable. In a compounding alert dated June 7, 2019, the FDA described seven patients at one outpatient clinic who were injected with L-glutathione 200 mg/mL, the same substance at the same concentration, and developed nausea, vomiting, lightheadedness, chills, body aches and sneezing within minutes, one of them hospitalized. FDA testing found bacterial endotoxin in the powder at up to five times the appropriate limit. The cause was that the powder was labeled “Caution: Dietary Supplement” and had been used to compound a sterile injectable anyway. Seven years later, the same substance, the same concentration and the same failure mode. FDA compounding alert, June 7, 2019.

That is the sentence to carry off this page. The recurring danger in injectable peptides is not that the molecule turns out to be toxic. It is that something which was never manufactured to be injected gets injected, and endotoxin is what a reader feels within minutes when it does. It is the same finding as the sterility column in the table above, arriving from the other direction.

Has anyone actually been hospitalised by peptides?

Yes. Two clusters are on the public record, in two countries, and between them 9 people were hospitalised. Neither involved a peptide dispensed against a prescription by a licensed pharmacy. Both involved product bought or handed over outside that route. And in neither case did anyone ever obtain the product and test it, which is the part that decides how much either record can actually prove. What follows is each cluster with the count that was confirmed, the authority that confirmed it, and, where a bigger number circulated, why it never became a fact.

What the regulator did about Las Vegas, eight months later

The injections were in July 2025. The Nevada State Board of Pharmacy issued fines the month before ProPublica reported them, reported March 13, 2026, totalling $35,000 across 4 parties. Every one of them is a licensing or distribution violation. None of them is a finding about what the product did, because the board never had the product.

• Dr Kent Holtorf, $10,000. The doctor who ran the booth, whose anti-aging practice is based in El Segundo, California, fined for practising in Nevada without a state licence.
• Han Bao Nguyen, $10,000. A pharmacist at Holtorf's practice, licensed in California, fined for the same unlicensed-practice violation, having mixed the peptides for both women and administered the serums to one of them.
• Michael McNeal, $5,000. An integrative health coach and director of education at Integrative Peptides, a company founded by Holtorf, who the board says does not appear to hold any health care licences, fined for prescribing or recommending a peptide cocktail to one of the women.
• Forgotten Formula, $10,000. A Texas-based private membership association, fined for unlicensed wholesaling of drugs, for mailing the peptides into Nevada.

What the public was shown instead

Set that record against the coverage. On August 16, 2026, The Whole Story with Anderson Cooper premiered Stacked: Inside the Enhancement Craze, an hour with CNN International Correspondent Isobel Yeungon the enhancement economy. Peptides are in it by name: the network's own announcement says Yeung meets influencers promoting testosterone, steroids and peptides, and speaks with athletes embracing the movement. Yeung travels from Texas to Abu Dhabi to Las Vegas, including reporting from inside the Enhanced Games. That is the frame most people now have for this category, and it arrived in the same month as the seventh Victorian hospitalisation. The two rarely travel together, which is the gap this section exists to close.

If a peptide hurts you, where does that fact go?

Everything above this line is a record of harm somebody already noticed. That raises the question underneath it, which is the one we could not find answered anywhere: when the next person is hurt, what system is supposed to write it down? We read the four that could plausibly apply, at source, on 2026-08-29. The answer is not that nobody is watching. It is stranger than that. Two of these systems will take a report from literally anyone, and none of them has a single person under a legal duty to file one about the peptides most readers are actually buying.

The doors that are open

Start here, because a page that leads with the gap would leave you thinking there is nowhere to go, and there is.

Neither of those has a mandatory reporter behind it. Both depend entirely on the injured person, or the clinician standing over them, choosing to make contact. That is a real cost of buying outside the system and it is worth naming plainly: people who bought a vial they knew was not meant for them are the least likely group in medicine to phone a federal agency about it afterwards.

The systems that require reports, and why they miss

US law does compel adverse-event reporting. It just compels it from two specific parties, and this supply chain contains neither.

The duty follows the pharmacy, not the risk

This is the sharpest thing we found, and it took reading both compounding sections end to end rather than trusting a summary of either. Whether anyone owes the FDA a report about a compounded peptide depends on which kind of facility made it, and not at all on how dangerous it is.

The compounding route this site keeps pointing readers toward, the one that runs through a state-licensed pharmacy under 503A, is on the side of that line with no reporting requirement. We are not going to soften that, because it is the honest cost of the advice. The supervised route is still the safer route, for the reasons the recall table above lays out, and it is also the route that generates no federal safety signal when it goes wrong.

The one thing on the label you can check yourself

Almost everything on this page is something a reader cannot verify without a laboratory. Purity, potency, sterility and what else is in the vial are all closed to you. One thing is not. If a vial or its paperwork names the pharmacy that supposedly compounded it, that name is a claim about a public record, and public records can be searched. It takes about ninety seconds and it is the only check on this page you can finish yourself.

Read the result the right way round, because the check is only powerful in one direction: a failed lookup tells you a great deal and a clean one tells you very little.

Why bother, when a recall would presumably warn you? Because it would not. We queried the openFDA drug enforcement database on September 7, 2026for every recall whose stated reason contains the word “counterfeit”, and it returned no matches found: 0 records, HTTP 404. Read that the way this page reads every other zero. A recall is an action a regulator takes against a manufacturer it regulates, and whoever counterfeits a label sits outside that chain entirely, so the database structurally cannot see them. The absence of counterfeit recalls is a fact about the register, not about how common counterfeits are. It does mean you cannot wait to be warned.

Sources for this section were read on September 7, 2026. The FDA pages are served to automated readers through an Internet Archive snapshot rather than live, because fda.gov blocks them, so the compounding questions and answers is quoted as of its September 2, 2026snapshot. What is quoted is how oversight is divided by statute rather than anything that changes week to week. We are not passing on the adverse-event counts that news coverage carried alongside this story: we did not take them to the FDA’s own reporting database, so they are not stated here as fact.

What does safe, supervised access look like?

Supervised access means a licensed clinician, a diagnosis or legitimate indication, and a peptide dispensed by a regulated compounding pharmacy or as an FDA-approved drug, not a vial mailed from an anonymous vendor. Today that route only exists for the legally-accessible set: FDA-approved drugs and peptides compoundable under 503A, like sermorelin. For the peptides currently under FDA review, there is no compliant supervised route until the ruling clarifies their status, which is the honest reason we attach no “where to buy” link to them.

See what the FDA is deciding on the July 2026 PCAC tracker, and the full legal picture on are peptides legal? For specific reactions and interactions, see peptide side effects.

Are peptides safe for weight loss?