Peptide half-life: what it means, and what it does not

What a half-life actually measures

A half-life is the time it takes for the amount of a drug in your blood to fall by half. It measures the molecule leaving your circulation, nothing else. It does not measure how long the drug works, how strong it is, or how safe it is, and reading it as any of those is where peptide charts go wrong.

Start with the plain version. Pharmacokinetics (what your body does to a drug, as opposed to what the drug does to your body) describes a molecule's journey in and out of you. The single number people quote from that journey is the elimination half-life: the time it takes for the amount circulating in your blood to fall by half.

The word half is doing a lot of work, because halving repeats. One half-life leaves half. Two leaves a quarter. Three leaves an eighth. By the fifth, what remains is a few percent of where you started, which is why pharmacologists treat five half-lives as the rough point where a drug stops mattering. That is arithmetic rather than biology, and it is the part of this subject you can check for yourself.

Watch it work on a real label. The prescribing information for Ozempic states that with an elimination half-life of approximately 1 week, semaglutide will be present in the circulation for about 5 weeks after the last dose [1]. Five weeks is five weekly half-lives, and the manufacturer is telling prescribers the same thing the arithmetic does: the drug does not stop when the injections stop.

Notice what that sentence is about. It is about molecules in plasma (the liquid part of your blood, the part a blood draw measures). It is not about whether you still feel anything, whether the receptor is still occupied, or whether a tissue somewhere is still responding. Those are different questions with different answers, and the rest of this article is mostly about how far apart the answers can be.

Your own peptides disappear in minutes

The peptides your body makes are built to be short-lived, because they are signals rather than supplies. Native GLP-1 is chopped at one end by an enzyme almost immediately after release. In the classic human study, most of what was injected had already been converted to an inactive fragment half an hour later.

Your gut releases GLP-1 (glucagon-like peptide-1, the hormone that tells your pancreas a meal has arrived) every time you eat. It is the natural molecule semaglutide was designed to imitate [4], and one of the two that tirzepatide acts on: tirzepatide is based on the GIP sequence, the backbone of a different gut hormone (glucose-dependent insulinotropic polypeptide), and selectively binds to and activates both the GIP and GLP-1 receptors [5]. On its own, GLP-1 is almost comically fragile.

Deacon and colleagues showed why in 1995, in a study that is still the reference point [3]. They gave GLP-1 to healthy volunteers and to patients with diabetes, then used assays that could tell the intact hormone apart from its chopped-up remains. A metabolite is what is left after the body has broken a molecule down, and here the metabolite is the hormone with two amino acids clipped off the front, which switches it off. The enzyme responsible is DPP-4, and being stabilised against it is one of the two things a long-acting GLP-1 drug has to manage [1]. Thirty minutes after a subcutaneous injection (under the skin, the way most peptides are given), that inactive fragment accounted for 88.5 percent of the rise in diabetic patients and 78.4 percent in healthy subjects, eight people in each group [3]. Given straight into a vein, intact GLP-1 made up only 19.9 percent of the signal in healthy subjects and 25.0 percent in the patients [3].

Enzymes are only half the story. Your kidneys filter small molecules out of the blood continuously, and peptides are small. A recent analysis of peptide pharmacokinetics puts the human filtration rate at about 0.11 litres per kilogram per hour, which by itself corresponds to a half-life of roughly 1.6 hours for a peptide that is not bound to plasma proteins [2]. That is the floor. Any unmodified peptide floating free in your blood is heading for that number or faster, and the enzymes make it faster.

How a once-weekly injection gets built

Long-acting peptides are not naturally long-acting. Drug designers bolt a fatty acid onto the chain so it clings to albumin, the most abundant protein in blood. A peptide riding on albumin is hidden from the kidneys and from the enzymes, and the half-life stretches from hours to days or weeks.

Albumin is the workhorse protein of your bloodstream, and it is far too big for your kidneys to throw away. The entire trick behind weekly peptide injections is to make a peptide hold onto it.

The semaglutide discovery paper says so outright: the aim was to design a once weekly GLP-1 analogue by increasing albumin affinity and securing full stability against metabolic degradation [4]. The molecule that came out has two amino acid substitutions compared with human GLP-1 and is derivatized at lysine 26, the attachment point for the fatty acid that does the clinging [4]. The paper reports a plasma half-life of 46.1 hours after an intravenous dose in mini-pigs [4], and it is worth being precise about that, because the figure gets quoted online as though it were a human result. It is a pig result. The human number is on the label: approximately 1 week, with semaglutide extensively bound to plasma albumin at over 99 percent, and the label names albumin binding as the principal mechanism of protraction, working by decreasing renal clearance and protecting the peptide from metabolic degradation [1].

Tirzepatide is the same idea with a longer tail. Its label describes a C20 fatty diacid that enables albumin binding and prolongs the half-life, reports the drug as 99 percent bound to plasma albumin, and gives an elimination half-life of approximately 5 days, which it says is what enables once-weekly dosing [5]. In the renal study that supported its approval, investigators kept drawing blood for 648 hours after a single injection, which is most of a month [10]. That is not caution for its own sake. It is what you have to do to characterise a molecule that takes five days to halve.

The floor, the ceiling, and why exenatide is twice daily

Peptide half-lives are penned in between two hard limits. Without albumin binding a peptide is gone within a few hours, which is why exenatide is injected twice a day. With extreme binding the peptide can only last as long as albumin itself does, and albumin turns over in about three weeks.

Exenatide is the control experiment nobody had to run, because it was already on the market. It is a GLP-1 receptor agonist, and the peptide pharmacokinetics analysis groups it with the unconjugated peptides rather than with the fatty-acid conjugated ones that tirzepatide, liraglutide and semaglutide belong to [2]. Its label reads accordingly: mean apparent clearance in humans of 9.1 litres per hour and a mean terminal half-life of 2.4 hours, with concentrations measurable for approximately 10 hours after a dose [6]. That is why it is injected twice a day and semaglutide is injected once a week. The receptor is the same. The staying power is not. Liraglutide sits between the two: like semaglutide it carries a fatty acid, and its plasma half-life is 13 hours after a subcutaneous dose [2], which buys once-daily dosing rather than once-weekly.

At the other end there is a ceiling, and it is set by albumin's own lifespan. Albumin circulates with a half-life of approximately 21 days, which puts a physiological limit on how far a peptide's half-life can be pushed by binding to it harder [2]. Simulations in the same analysis show the upper boundary approaching approximately 21 days at very high binding, and, more usefully for reading a chart, that any extension beyond approximately 1 week requires the peptide to be unusually resistant to metabolism rather than merely sticky [2]. A weekly peptide is already near the practical edge of what albumin binding alone buys you.

There is a third number in the same family that explains a detail people find odd. Peptides cross membranes poorly, so they stay largely in the fluid outside your cells, around 18 litres in an adult [2]. That is why their volumes of distribution (roughly, how widely a drug spreads through the body) sit in a band the same analysis calls markedly narrower than that observed for small molecules [2]: approximately 12.5 litres for semaglutide [1], approximately 10.3 litres for tirzepatide [5], and 28.3 litres for exenatide [6]. The exenatide figure is the one that lands above extracellular fluid, and it is worth knowing why before reading it as tissue uptake: all three were measured after a subcutaneous injection, which makes them apparent volumes, so whatever fraction of the dose never reached the blood inflates them [2]. A peptide is not soaking into your tissues the way a small-molecule drug does. It is mostly in the water it was injected into.

A half-life is about the molecule, not the effect

The most common mistake is reading a half-life as a duration of action. BPC-157 is the clearest case: the intact peptide has a half-life of under half an hour in every species measured, including the only two humans whose blood levels have been published, while the effects attributed to it are described as running on for hours to days.

BPC-157 got its first proper pharmacokinetic study in 2022, and the numbers are not ambiguous [7]. After a single intravenous dose the average elimination half-life in rats was 15.2 minutes, and in beagle dogs it was 5.27 minutes [7]. Across every dose and both species the authors summarise it as less than 30 minutes [7]. The only human pharmacokinetic data that exist come from a two-subject pilot in which a 58-year-old man and a 68-year-old woman received intravenous infusions of 10 mg and 20 mg on consecutive days: half-life under 30 minutes again, with plasma concentrations back to baseline within 24 hours [8].

Now hold that against what the compound is sold on. A 2026 review of BPC-157's development sets the two side by side: a plasma half-life of under 30 minutes, confirmed preclinically and in that preliminary two-subject human pilot, contrasts with prolonged biological effects lasting hours to days, a disconnect with significant implications for dosing strategy and formulation design [8]. Whatever is or is not happening in those later days, the blood level is not what is happening. A short half-life does not rule out a long effect, and a long half-life does not guarantee one.

The 2022 study contains the cleanest demonstration that a half-life is a measurement rather than a property. When the researchers tagged BPC-157 with tritium and, after an intramuscular injection, tracked total radioactivity instead of the intact peptide, the half-life came out at 102 hours [7]. Nothing about the molecule changed. The label simply followed the wreckage: the peptide is rapidly broken into small fragments and then into single amino acids, which enter your ordinary amino acid metabolism and go on circulating as food [7]. 15.2 minutes and 102 hours are both correct answers, to two different questions.

How to read a half-life you find online

Three questions turn a number in a chart back into evidence. Which species was it measured in, by which route, and what exactly was being counted. A figure that cannot answer all three is not a fact about the peptide, and for most compounds sold as research chemicals there is no human answer at all.

Start with species. Semaglutide's famous 46.1 hours is a mini-pig [4], BPC-157's 15.2 minutes is a rat, and its 5.27 minutes is a dog [7]. None of those is a human number, and peptide half-lives scale with body size rather than transferring across [2].

Then ask about route. Everything quoted above for BPC-157 was measured after an injection into a vein or a muscle [7] [8]. Subcutaneous injection, which is how people actually use it, has not been characterised [8]. Absorption from under the skin is its own process with its own speed, and for a fast-clearing peptide it can end up controlling the shape of the curve entirely.

Then ask what was counted, which is the tritium lesson. Intact peptide [7], total radioactivity [7], and an assay that cannot tell a hormone from its inactive fragment [3] will give you three different answers from the same blood.

Finally, ask whether there is a number at all. More than 80 peptide drugs have reached the market since insulin, covering diabetes, cancer, osteoporosis, multiple sclerosis, HIV infection and chronic pain [9], and those are the ones with labels, because a regulator required the work. BPC-157 is the instructive counterexample: it has been circulating in the research-peptide trade for years, and the 2022 paper describes itself as the first analysis of the pharmacokinetics of BPC-157 [7]. When a vendor chart lists a tidy half-life for a compound nobody has studied in people, that number came from somewhere, and it is worth asking where.