Do Peptides Actually Work? Three Questions, Three Answers

Ask whether peptides actually work and you will get a confident answer either way. Both answers are usually wrong, because the question is being asked about three different things at once.

One group of peptides has been through trials with thousands of participants. Another group sits in skincare, with a small and formulation-specific record behind it. A third group sells as research material and has barely been tested in people at all. Sorting those apart is most of the work.

A disclosure belongs here in the body, not buried in a footer. Helix Bio Chem sells research-grade peptides, and several compounds graded below are in the catalogue. Everything here assesses published literature, and the catalogue shaped no verdict in it. Where the evidence for something we sell turns out to be thin, this article says so.

The short answer

Do peptides actually work? Not as one category with one answer. A few are approved medicines backed by trials in thousands of people. Most compounds sold as research peptides have little or no human testing, and some have none. Topical cosmetic peptides form a third, separate evidence base with weaker and narrower support.

Where each compound actually stands

Evidence type describes what kind of study exists. Confidence describes how much weight that study can carry for the claim people make about the compound.

Approved medicines

Investigational

Sold as research compounds

Different evidence base entirely

Three different questions get asked as one

Look at who ranks for this question and the confusion shows itself. A dermatology clinic, a skincare brand, a hospital system and a weight-loss service all compete for the same search. They are answering about different molecules.

The three groups barely overlap at all. Approved peptide medicines are prescription drugs that went through randomised trials and carry an approved indication. Topical cosmetic peptides sit in serums and creams, where the question is whether anything crosses the skin. Research peptides sell in vials labelled for laboratory use, and most of them have never been given to a person in a published study.

Our evidence grading methodology sets out how we handle each kind of study. The short version is that the question being asked decides which source counts. Chemical identity needs analytical work to settle. Efficacy needs a trial in the population you care about.

Readers who came here about serums will get more from our article on peptides for skin, which separates the oral and topical records properly. The rest of this page is about the injectable and research side.

The peptides with real human trials

Start with what the strong end looks like, because it sets the standard everything else gets measured against.

STEP 1 enrolled 1,961 adults and ran 68 weeks. Semaglutide reduced body weight by 14.9% against 2.4% on placebo (PMID 33567185). SURMOUNT-1 enrolled 2,539 and ran 72 weeks, with tirzepatide reaching 20.9% at its top dose against 3.1% (PMID 35658024). Both trials published full adverse event tables. Neither enrolled people with diabetes, so the diabetes indications rest on different programmes.

Retatrutide now belongs in the same conversation about evidence, though not about availability. TRIUMPH-1 randomised 2,339 adults over 80 weeks and reported a 25.0% fall in body weight at 12 mg against 3.9% on placebo (PMID 42814954). The compound remains investigational for now. Eli Lilly has not filed a marketing application. A vial sold as retatrutide today is an unapproved drug whatever the trial showed.

Tesamorelin is the most instructive entry in this group, and almost nobody cites it accurately. Its pooled phase 3 analysis covered 806 adults and measured a 15.4% treatment effect on visceral fat (PMID 20554713). Read the population before the number. Every pivotal trial enrolled adults with HIV and lipodystrophy, a specific pattern of fat redistribution. Subcutaneous fat did not change, and neither did limb fat. Nothing in the approval package measured strength, recovery or lean mass, and the label states the product is weight neutral.

Elamipretide makes a similar point from a different angle. It is approved, and it is approved to improve muscle strength, which sounds like it should settle an argument. The indication covers Barth syndrome in patients weighing at least 30 kg, under accelerated approval based on an intermediate endpoint. An ultra-rare mitochondrial disorder tells you nothing about a healthy adult's shoulder.

The peptides almost nobody has tested in people

Now the other end, where the gap between reputation and evidence is widest.

BPC-157 carries the biggest reputation of any compound in this group. A systematic review registered in advance searched Embase, PubMed, the Cochrane Library and ClinicalTrials.gov to June 2026. It found the entire human record to be three studies in 30 people (PMID 42885452). All three studies ran without a control group. Each graded poor on methodological quality, with overall certainty rated very low. The reviewers concluded the compound should not be used clinically outside regulated trials. Worth noting too that all three studies share one first author and one private clinic, which is a structural limit on its own. Our review of what BPC-157 research shows covers the structure and proposed mechanisms.

TB-500 has a problem before evidence even enters the picture. Published analysis of a product sold under that name identified Ac-LKKTETQ, an acetylated seven-residue fragment (PMID 22962027). Thymosin beta-4, the molecule TB-500 is usually described as, runs 43 residues (PMID 17947592). Those are different molecules, roughly six-fold apart in size. The human trials that do exist in this area used full-length thymosin beta-4, topically, for eye and wound indications. No human study of the seven-residue fragment was located. Our article on TB-500 and thymosin beta-4 identity works through the analytical detail.

KPV is a genuine tripeptide and the anti-inflammatory signal in rodent colitis models is real. Two gaps matter more than the rest. The receptor it acts through has not been identified, and a 2010 review states the signalling mechanism is unknown (PMID 21222263). There is no human trial of KPV for any indication by any route. The animal studies with the clearest results used a hydrogel or nanoparticle delivery system. One of those papers notes the free peptide is unstable when given rectally. A plain vial has none of that engineering.

Growth hormone secretagogues deserve their own treatment, and they have it. Our review of growth hormone secretagogues in muscle research sorts the human studies by exact molecule and by endpoint. The detail belongs in that article, not this one. In short, these compounds raise growth hormone and IGF-1 dependably, and the trials measuring what happens next tell a different story. More on why that gap matters below.

MOTS-c illustrates a trap worth naming. Human studies of MOTS-c measured how much of it people already had circulating, and found lower levels in some groups. That is an association in a population, not a test of giving it to anyone. The first interventional trial of MOTS-c itself, registered as NCT07505745, started recruiting in February 2026 at a single site, with results not expected before 2027. A trial being run shows that someone is testing a hypothesis properly. It does not show the hypothesis is right.

AOD-9604 is the most interesting failure in this group, and the clearest case for reading mechanisms carefully. Its mechanism is real and reproducible. It does not compete for the growth hormone receptor and does not drive cell proliferation through it, unlike growth hormone itself. In obese mice it raised fat oxidation and lipolysis and reduced weight gain, without the blood glucose rise growth hormone produced (PMID 11673763). Then a 24-week obesity programme in roughly 500 participants missed its primary weight-loss endpoint. That failure reaches us through a 2026 review reading sponsor-reported outcomes, not a published trial paper (PMID 42395176). The sourcing is itself a limitation worth stating. A plausible mechanism and a working drug are separate achievements.

GHK-Cu has the widest gap between its research profile and its human record. A 2026 systematic review screened 1,247 records and included 64 studies (PMID 42787770). Twenty-four of those were reviews, not primary research. Eleven were human clinical studies, and the review reports that eight of the eleven did not test GHK-Cu as the experimental intervention. Three human clinical studies tested the compound itself. The review found no published full-text randomised trial of topical GHK-Cu delivered by microneedle array. Its authors named that clinical evidence gap as the field's main problem. For injection, hair growth and systemic tissue repair, the human record is empty. Our GHK-Cu research overview covers the mechanistic work in more depth.

Why "it worked for me" is weak evidence here

Personal experience feels like the strongest evidence available. In this particular situation it is among the weakest, and three mechanisms explain why.

Placebo response in joint pain is large and well measured. A meta-analysis covering 198 trials compared 193 placebo groups, holding 16,364 patients, against 14 untreated control groups holding 1,167 (PMID 18541604). Placebo relieved pain with an effect size of 0.51. The untreated control groups registered just 0.03. That same analysis found the placebo response grew significantly larger for placebos given by injection or needle. Read that alongside the BPC-157 record. Its main human study asked people with knee pain how they felt after injections, with no control group and no blinding.

A caution belongs with that figure, because the placebo literature cuts both ways. One Cochrane review covered 234 trials across 60 conditions, with outcome data from 202 of them (PMID 20091554). It found placebo effects on pain to be modest, with an effect size near 0.28. Its authors also noted how hard it is to separate a real placebo effect from biased reporting. Large improvements in uncontrolled case series are better explained by reporting bias and natural history than by a dramatic placebo effect.

Regression to the mean does more work than most people expect. Symptoms that fluctuate tend to drift back toward their average, and an unusually bad measurement is usually followed by a better one. The methodological account of this phenomenon notes that the effect grows when you select people on a baseline value and then follow them (PMID 15333621). Picture when anyone starts a recovery compound. They start when the knee is at its worst. The structure of that decision builds improvement into the next measurement before anything pharmacological happens.

Polypharmacy finishes off the attribution problem. People using these compounds often use several at once, sometimes alongside anabolic steroids, thyroid hormone or stimulants. Our peptide stacking guide explains why combining agents makes attribution harder, not easier. If three things change and one variable moves, nothing in that observation identifies which agent moved it.

None of this means nobody improves. It means an uncontrolled improvement cannot tell you what caused it, which is the entire reason controlled trials exist.

Why a result in a rat is not a result in a person

Most of the enthusiasm around research peptides traces back to animal work, and the animal work is often genuinely good.

A PRISMA scoping review published in 2026 looked at BPC-157, TB-500, CJC-1295, MK-677, ipamorelin and GHK-Cu, crossed against musculoskeletal tissue (PMID 42578445). Across the publications it identified, 67% used preclinical animal models. Human clinical studies amounted to a handful, most lacking robust controls or rigorous design. The authors concluded these supplements should not be recommended as a replacement or an addition to existing orthopaedic care.

Species is only the first gap. Dose rarely translates, because a milligram per kilogram in a mouse is not a human dose. Route changes everything, and a compound injected into a rat's tissue behaves differently from one injected under human skin. Endpoint substitution does the rest of the damage, because a histology slide showing more blood vessels is not a tendon that works better.

The growth hormone literature shows what happens when somebody actually closes that loop. A systematic review of 27 study samples, covering 303 people who received growth hormone, found lean body mass rose by 2.1 kg (PMID 18347346). Strength and exercise capacity did not improve. Lactate during exercise ran higher in two of the three studies that measured it, and the treated participants reported more soft tissue swelling and fatigue. The authors wrote that claims about growth hormone enhancing physical performance are not supported by the literature.

MK-677 makes the same point from the secretagogue side, which is why it keeps appearing in these discussions despite not being a peptide. A two-year randomised trial in 65 healthy older adults raised fat-free mass by 1.1 kg against a 0.5 kg loss on placebo (PMID 18981485). Strength and function did not change. A separate phase 2b trial in 123 people recovering from hip fracture stopped early over a congestive heart failure signal. Four cases appeared in 62 treated patients against one in 61 on placebo. The authors described the safety profile as unfavourable in that population (PMID 21067829). Four against one in groups that size is not a demonstrated effect, and it was enough to halt a trial.

Compounded, approved, investigational and research grade are four different things

These four words get used interchangeably, and each one means something specific.

An approved drug has been through trials and carries an FDA-approved indication. Investigational means still in trials with no approval, which is where retatrutide sits. A compounded product is made by a pharmacy for a particular patient, and it is not the product the pivotal trials studied. Research-grade material sells for laboratory use and has no approval of any kind.

The compounded category causes the most confusion, because the molecule name matches an approved drug. Semaglutide and tirzepatide are no longer in shortage, and the shortage-based pathway that allowed widespread compounding closed during 2025. What remains is narrower, and our article on whether compounded semaglutide and tirzepatide are still legal walks through the current position. The evidence point is simpler than the legal one. Trial results belong to the product that was tested, at the dose and purity that trial used.

What is actually in the vial

Everything above assumes the compound in the container matches its label. Published analytical work says that assumption needs checking, and this applies to the unregulated supply chain as a category, not to any particular seller.

Belgian public-health researchers profiled the ten most frequently encountered falsified peptide drugs on their market, bought from three suspected illegal internet pharmacies (PMID 30029448). Purity for cysteine-containing peptides ran between 5% and 75%. The analysts also found arsenic and lead. One of the samples carried lead contamination. Several carried elemental impurities at up to ten times the ICH toxicity limit for injectable drugs. Arsenic speciation confirmed all of it sat in the more toxic inorganic form.

A 2024 study bought semaglutide from six illegal online pharmacies without a prescription (PMID 39509151). Three vials arrived and three prefilled pen orders turned out to be non-delivery scams. Measured semaglutide content exceeded the labelled amount by 28.56% to 38.69%, so the direction of error was over, not under. Purity measured between 7.7% and 14.37% against a 99% claim on the label. Endotoxin turned up in every single sample.

Our article on peptide purity testing covers the contamination data in more depth.

Where the regulation actually stands

Dated statements matter here, because this area moved several times during 2026.

FDA's Pharmacy Compounding Advisory Committee met on 23 and 24 July 2026 and considered seven peptides nominated for the 503A bulk drug substances list. Each substance received two votes, one for its free base and one for its acetate form. FDA's own briefing documents proposed that none of the fourteen forms be included. Our explainer on the July 2026 FDA peptide vote covers what the committee recommended and what it did not.

A committee recommendation carries no legal effect. Adding a substance to the 503A list happens only through notice-and-comment rulemaking, as our explainer on the 503A bulks list describes. As of 22 October 2026, none of those seven peptides sits on the list, none is FDA-approved, and compounding them is not lawful. Our list of FDA-approved peptides covers what has actually cleared approval.

What would change these verdicts

Every grade above is provisional, and the specific thing that would move each one is worth naming.

BPC-157 needs a randomised controlled trial with an objective healing endpoint. Thirty people in uncontrolled studies cannot answer the question whatever the results look like. TB-500 needs a basic identity study first, establishing what sits in products sold under that name. Human work on whatever molecule that turns out to be comes after. KPV needs a phase 1 trial, and a delivery system resembling the one that worked in rodents. MOTS-c has its answer coming, since NCT07505745 is designed to test the hypothesis properly and should report after 2027.

GHK-Cu needs the trial its own systematic review called for. That means an adequately powered randomised study of a topical formulation, using a delivery method that crosses skin. The secretagogues need trials measuring function, not hormone levels, and the growth hormone literature suggests what those might find.

Retatrutide moves in the opposite direction. Its evidence is strong and its availability is nil, so the thing that would change its entry is regulatory, not scientific.