All compounds

Gray-market GLP-1s

Also sold as research-grade semaglutide, compounded semaglutide, research peptide tirzepatide, retatrutide. GLP-1 receptor agonists obtained outside the regulated supply chain.

Serious risk (4 of 5)

Semaglutide and tirzepatide are among the best-studied drugs of the last decade — and the danger in buying them as research peptides has almost nothing to do with the molecule and almost everything to do with a person doing arithmetic with a syringe at their kitchen table.

What it is

This page is not about semaglutide or tirzepatide. Those are approved medicines with large, high-quality trial programmes behind them, and if you have been prescribed one, this page is not about what you are taking.

This page is about the parallel market: vials of powder sold online as “research grade” semaglutide, tirzepatide or retatrutide, sold by peptide vendors under a research-use-only label, and bought by people who intend to inject them. Prices are a fraction of the pharmacy cost, availability does not depend on a prescriber, and the whole thing is one search away.

The reason people do it is entirely comprehensible. These drugs work extremely well for a condition that carries real medical consequences and real social cost. They are expensive, frequently not covered by insurance, and for stretches of the last few years simply unavailable. Telling someone in that position that they should have gone through proper channels is not advice; it is a description of a door that was closed.

So the useful thing this page can do is be specific about where the danger actually sits — because it is not where most people assume, and the difference is actionable.

The molecule is the safest part of this. The vial and the syringe are the dangerous parts.

What it actually does

GLP-1 receptor agonists mimic a gut hormone released after eating. They slow stomach emptying, act on appetite regulation in the brain, and improve insulin response. Tirzepatide acts at a second receptor as well.

The effect is that hunger — including the intrusive, constant kind that people describe as background noise they cannot turn off — substantially quietens. Weight comes off, and it keeps coming off in a way that diet and exercise alone reliably fail to produce for most people. In people with type 2 diabetes, blood sugar control improves markedly. There are cardiovascular benefits demonstrated in large outcome trials.

This is one of the genuinely effective drug classes of the last twenty years. Nothing on this page disputes that, and any page that opened by implying these drugs are dubious would deserve to be closed.

What the evidence actually shows

The evidence splits cleanly in two, and keeping the halves separate is the whole point.

On the drugs themselves: extensive and high quality. Large randomised trials, cardiovascular outcome trials, years of follow-up, and post-marketing surveillance across millions of patients. The side effects are well characterised — nausea, vomiting, constipation, gallbladder problems, pancreatitis as an uncommon but real event, and a warning about medullary thyroid carcinoma carried over from rodent studies. You can look all of this up, and a prescriber monitors for it.

On what is actually in an unregulated vial: this is the problem.

A 2024 market-surveillance study bought semaglutide products from online sellers operating without prescription requirements and analysed both the sellers and the products. The findings included illegitimate operations, products whose content did not match their claims, and quality failures of exactly the kind you cannot detect by looking at a vial.

Separately, analysis of custom synthetic peptides has found undeclared constituents present. A powder that looks correct tells you nothing about its identity, its strength, or what else is in it.

And on the step that actually harms people: a poison-control case series documented administration errors with compounded semaglutide — people receiving many times the amount intended, presenting with severe vomiting, dehydration and hypoglycaemia, some requiring hospital treatment.

That last finding is the one that should change behaviour. The harm was not caused by the drug being fake or contaminated. It was caused by arithmetic. Somebody had to convert a quantity of powder into a volume of liquid into marks on an insulin syringe, and got it wrong by a factor of ten. That is not an exotic failure. It is the single most predictable error in the entire process, and it is the one the regulated supply chain exists to make impossible — a pharmacy product arrives at a known strength in a device that measures it for you.

The risk profile

Serious risk (4 of 5) — Documented harm in humans that is severe, common, or hard to see coming.

What going wrong looks like from the inside

There are two ways this goes wrong, and they run on completely different timescales.

The fast one is an overdose, and it happens on your first attempt more often than later. You have a vial of powder, a bottle of liquid, and an insulin syringe marked in units that have nothing to do with the amount of drug you are trying to give. Somewhere between those three things is a conversion, and it is the kind of conversion that is easy to get wrong by a factor of ten — a misplaced decimal point, a number read off the wrong post, a syringe with different markings from the one in the guide you were following.

If you get it wrong upward, you will know within hours. It starts as nausea and quickly becomes vomiting you cannot stop. You cannot keep water down. If you are also on diabetes medication, your blood sugar can fall dangerously. People have needed intravenous fluids and admission. There is no antidote — these drugs are long-acting, so once it is in you, it is in you for days, and treatment is supportive.

The point to internalise: this failure happens at the kitchen table, not in your body. It is a measurement error, which means it is preventable by treating the measuring step as the dangerous step. If you are going to do this, do the arithmetic twice, on paper, on a different day, and have another person check it — that single habit removes most of the documented harm on this page.

The slow one is that nobody is watching. Prescribed, these drugs come with monitoring: someone checks how fast you are losing weight, notices if you have stopped eating protein, catches gallbladder symptoms, knows to take severe persistent abdominal pain seriously as possible pancreatitis, and reviews your other medication. None of that exists when the drug arrives in the post.

The symptoms worth knowing by name, because there is no one else to know them for you: severe abdominal pain radiating to the back, especially with vomiting — that is the pancreatitis presentation and it needs an emergency department the same day. Pain in the upper right abdomen, particularly after eating, with fever or yellowing skin — gallbladder. Persistent vomiting with an inability to keep fluids down — dehydration, which becomes dangerous faster than people expect.

And a specific, under-appreciated one: if you need surgery or a procedure involving sedation, tell them you are on a GLP-1 agonist. These drugs slow stomach emptying, which means you may have food in your stomach after fasting normally, and that creates a real aspiration risk under anaesthesia. Anaesthetists now ask about this routinely. If you are using one they do not know about, the question does not get asked.

Interactions and combinations

Insulin and sulfonylureas are the interaction that causes acute harm. GLP-1 agonists lower blood glucose, and combined with diabetes medication that also lowers it, hypoglycaemia is a genuine risk. Anyone on diabetes medication needs that regimen reviewed by a prescriber — this is not optional, and it is the most common route to a serious event.

Anything taken by mouth is affected, because the stomach empties more slowly. That includes oral contraceptives, thyroid medication and antibiotics. A pharmacist will answer this without judgement and without needing to know where the drug came from.

Alcohol is worth mentioning for a practical reason: many people find their tolerance changes substantially on these drugs, and the combination with the nausea is unpleasant.

Stacking with other peptides from the same vendors multiplies the sterility and measurement risk without adding anything.

If you’re already using it

There is no withdrawal. Stopping is stopping, though appetite returns and weight regain is common — that is a property of the drug class rather than of the gray market.

The genuinely useful things:

Treat the measuring step as the dangerous step. Do the conversion twice, write it down, and have someone else check it. This is where the documented harm comes from.

Tell your doctor. This is the one that people avoid and the one that matters most. You do not have to explain how you obtained it, and no reasonable clinician will make it a confrontation. What it buys you is monitoring, an interaction check against everything else you take, and — critically — an accurate record if you ever end up needing surgery or an emergency department.

Tell an anaesthetist before any procedure, without exception.

Know the two emergencies: severe abdominal pain going through to your back, and vomiting you cannot stop. Both warrant same-day assessment rather than waiting it out.

Sterile technique matters. Clean hands, alcohol swab, never share anything, and treat redness, heat or swelling at an injection site with a fever as a reason to be seen.

If cost is why you are here, it is worth checking whether it still needs to be. Manufacturer direct-purchase programmes, changes in insurance coverage, and the arrival of cheaper approved options have moved a good deal in the last couple of years, and a real prescription removes every single risk on this page except the ones the drug itself carries.

This page has not yet been reviewed by a clinician. It was written from the published literature and every claim is cited, but no named medical professional has checked it. We say so here rather than let a missing byline read as an implicit one.

Last reviewed 7 September 2026. Found something wrong? Corrections are published, not silently edited.