Is Tirzepatide a Peptide? Where the Boundary Sits, and Why the FDA Drew It at Exactly Forty Amino Acids

Is Tirzepatide a Peptide? Where the Boundary Sits, and Why the FDA Drew It at Exactly Forty Amino Acids

Tirzepatide is 39 amino acids long. Add two more and it would be regulated under a different federal statute, by a different approval pathway, with biosimilars instead of generics. The line is at 40, it is deliberately arbitrary, and the FDA said so in writing when it drew it. Almost every "is X a peptide" question has two answers, one from chemistry and one from law, and the second one is the one with consequences.

Michael Phelps

Founder & Peptide Research Specialist, PrymaLab

Research reference · Last updated August 19, 2026 · ~19 min read

Chemistry Has No Line Here

Ask a biochemist where a peptide becomes a protein and you will get a shrug and a number that varies by who you asked.

The usual answer is somewhere around 50 amino acids, sometimes 100, sometimes framed as whether the chain folds into a stable tertiary structure rather than as a count at all. None of those is a law of nature. A 45-residue chain and a 55-residue chain are the same kind of molecule, built by the same bonds, made by the same machinery.

The folding criterion is the most defensible and it is still awkward. Plenty of short chains adopt stable structures and plenty of long ones are intrinsically disordered. Insulin folds properly at 51 residues; some proteins several times that length have no fixed structure at all.

So the honest chemical answer to "is X a peptide" is that it depends what you want the word to do, and that for most purposes the distinction does not carry any weight.

Which would be the end of the article, except that one organisation needed the word to do something specific and could not afford a shrug.

The FDA Drew One Anyway

On 23 March 2020 a final rule took effect defining the term biological product for regulatory purposes. In it, the FDA defines a protein as:

any alpha amino acid polymer with a specific, defined sequence that is greater than 40 amino acids in size^[1]

Which produces the complementary definition by subtraction. An amino acid polymer of 40 or fewer residues is not a protein, and continues to be regulated as a drug under the Food, Drug, and Cosmetic Act.^[1]

Two details in the rulemaking are worth having.

The FDA knew the number was arbitrary and said so. Commenters pointed out that proteins can be smaller than 40 amino acids. The agency did not dispute it. It kept the definition on the grounds that it establishes a scientifically reasonable, bright-line rule that provides regulatory clarity, reducing uncertainty for both the agency and applicants.^[1]

I find that unusually candid for a regulatory document. The line is not a claim about chemistry. It is an administrative decision to have a line at all, made openly, because the alternative was arguing about folding on a case-by-case basis forever.

Synthesis route does not matter. The FDA clarified that even chemically synthesised polypeptides over 40 amino acids fall within the definition of a biological product.^[1] So you cannot make something a drug rather than a biologic by building it on a synthesiser instead of in a cell. The count is the count.

What Sits on Either Side of the Number

The classification is not a label. It determines which statute governs the product, and the two regimes are different in ways that matter commercially.

A peptide at 40 or fewer residues is a drug under the FD&C Act. It goes through a new drug application, and after exclusivity expires it can face generic competition through an abbreviated pathway that rests on demonstrating the same active ingredient.

A protein above 40 is a biological product under the Public Health Service Act. It goes through a biologics licence application, and competitors arrive as biosimilars, which is a longer and more expensive route because you cannot simply assert chemical identity for a molecule whose properties depend on folding and glycosylation.

I have no idea whether anyone has ever designed a molecule with the number in mind, and I would be speculating if I claimed they had. What I can say is that tirzepatide is 39 amino acids and the line is at 40, and that is a narrow margin for something with that much riding on it.

Insulin, and the Day the Definition Took Effect

If you want to see what the boundary actually does, insulin is the clearest case available, because it crossed the line without changing.

Human insulin is 51 residues, an A chain of 21 and a B chain of 30 held together by disulfide bonds. Over 40. A protein by the definition.

For decades insulin was regulated as a drug, approved under the FD&C Act, because that was where it had always sat. On 23 March 2020, under a transition provision of the Biologics Price Competition and Innovation Act, insulin and a set of other previously-approved protein products were deemed to be licensed as biological products.^[1][2]

Nothing about insulin changed. Not the sequence, not the manufacturing, not the clinical use. The statute governing it changed, on a scheduled date, because of a residue count.

The practical point of the transition was to open insulin to biosimilar competition, which the drug pathway had not been delivering. Whether it worked is a separate argument I am not qualified to have. As an illustration of what the 40-residue line does, it is the cleanest one there is.

Semaglutide, and What Modification Does Not Change

Semaglutide is where people's intuitions break, and the reason is that it does not look like the tidy amino acid chain the word peptide suggests.

The backbone is 31 amino acids, based on the GLP-1 sequence. Then come the modifications that make it a weekly drug rather than a molecule with a two-minute half-life:

• A non-natural amino acid, aminoisobutyric acid, substituted at position 8 to block DPP-IV cleavage. Same enzyme problem described in the secretagogue comparison.
• A C18 fatty diacid attached through a spacer to the lysine at position 26, which binds albumin and keeps the molecule in circulation.

So is that a peptide? By the regulatory definition, yes, comfortably, because 31 is well under 40 and the definition counts amino acids in the polymer rather than asking what has been attached to them.

Biochemically it is a peptide with a substantial amount of non-peptide chemistry hanging off it, and by mass a meaningful share of the molecule is not amino acid at all. If your mental model of a peptide is a bare chain, semaglutide will not fit it, and the model is what needs updating rather than the classification.

The same applies across that class. Tirzepatide at 39 residues carries a similar lipid modification. Retatrutide, covered in the triple agonist comparison, is built the same way.

Two Words in the Definition That Do a Lot of Work

Read the FDA wording again and two qualifiers are doing more than the number is.

The definition covers any alpha amino acid polymer with a specific, defined sequence greater than 40 amino acids.^[1]

Alpha. That word restricts it to polymers of alpha amino acids, meaning the standard arrangement where the amino group sits on the carbon adjacent to the carboxyl. It leaves out beta-peptides, peptoids and other backbone-modified chains that chemists build precisely because they resist the enzymes that cut ordinary peptide bonds. A 45-residue beta-peptide is not a protein by this definition, whatever else it is.

Specific, defined sequence. This one has a consequence the collagen section below runs straight into. A hydrolysate has no specific defined sequence. Break a protein into fragments and you get a distribution of many sequences, so the material fails that clause regardless of how long any individual fragment happens to be. Collagen peptides are not proteins under this definition, and it is not because the fragments are short.

I would not have predicted either of those from the headline number, and both are the kind of detail that decides an argument if one ever arises about a particular product.

There is a related observation on the metabolic drugs that I want to state carefully. Native GIP is 42 amino acids. Tirzepatide, which is built on a GIP analogue backbone, is 39. So the natural hormone sits above the line and the drug derived from it sits below. I am not claiming anyone shortened it for that reason, and I have seen no evidence that they did. It is a fact about two numbers that I noticed while assembling the table and could not stop looking at.

Peptides Are Not Steroids

This question gets nearly ten thousand searches a month and the answer is short, so the section is short.

Steroids are built on a fused four-ring carbon skeleton, derived biosynthetically from cholesterol. Testosterone, cortisol, oestradiol, and the anabolic steroids all share that core. No amino acids anywhere.

Peptides are chains of amino acids joined by peptide bonds, with no ring system of that kind.

They also do different things. Steroids are small and lipophilic, so they cross the cell membrane and bind intracellular receptors that act as transcription factors. Most peptides bind receptors on the cell surface and signal through second messengers without entering the cell. That is the same divide covered in the SARMs and peptides comparison, since SARMs act on the same intracellular receptor steroids do.

The two categories share a marketplace and nothing else.

What Changes With Length, Analytically

The regulatory line is administrative. There is a separate, real gradient in how hard a molecule is to verify, and it tracks length closely enough to be worth setting out. This is the part of the article with a practical use.

Under about fifteen residues. Tandem mass spectrometry can sequence the molecule directly, reading it residue by residue from the fragmentation pattern. You get the mass and the residue order. KPV at three residues and BPC-157 at fifteen sit here. If a supplier of a short peptide cannot confirm sequence, that is a choice rather than a limitation of the technique.

Roughly fifteen to forty. Mass confirmation is straightforward, sequencing is harder, and the practical problem becomes deletion sequences: chains missing one residue from a coupling step that did not go to completion. A 30-mer missing one glycine differs by 57 daltons and can co-elute with the correct product on a routine gradient. Purity by a single-wavelength chromatographic method will not always separate them. VIP at 28 residues and semaglutide at 31 are here.

Above forty. Now folding matters, and no purity assay detects it. A chain with the correct sequence and the correct mass can be misfolded and inactive, and reversed-phase chromatography will not tell you. If the material was expressed in cells rather than synthesised, glycosylation enters as well, and glycosylation patterns vary with the host cell line. Size-exclusion chromatography becomes the method that earns its place, because it detects the aggregates these molecules form readily.

That gradient explains why the recombinant items discussed in the myostatin reference need different certificates from the fifteen-residue compounds that make up most of this catalogue. It also explains why the 40-residue line, arbitrary as the FDA admitted it to be, is not a stupid place to put it. Somewhere around there is where the analytical questions genuinely change.

Things in Peptide Catalogues That Are Not Peptides

Including ours, which is why I am writing this section rather than leaving it out.

A peptide catalogue is organised by what people buy together, not by chemical class. That is a reasonable commercial decision and it produces a category label that is not a chemical claim. Some examples from our own shelves:

• NAD+ is a dinucleotide of roughly 663 daltons, built from adenine, nicotinamide, two ribose sugars and two phosphates. No amino acids. The delivery problems that follow from its size and charge are covered in the NAD+ routes reference.
• Methylene blue is a phenothiazine dye. A small organic molecule, no relation.
• 5-amino-1MQ is a small molecule enzyme inhibitor.
• SARMs contain no amino acids at all, as covered in the comparison above.
• Tesofensine is a small molecule reuptake inhibitor.

Two genuine edge cases where the answer surprises people. Glutathione is a peptide, a tripeptide of glutamate, cysteine and glycine, though it is joined by an unusual gamma linkage at the glutamate rather than the standard alpha peptide bond. And cerebrolysin is a mixture of low molecular weight peptides and free amino acids derived from porcine brain tissue, which makes it peptide-containing rather than a peptide.

None of this is a criticism of how the industry organises itself. It is a warning against reasoning from the shelf to the chemistry. If a supplier tells you a product is a peptide, that is a statement about where it sits in their catalogue.

Collagen Peptides Are a Different Kind of Noun

Collagen peptides get more search volume than every research peptide combined, and they are a different sort of object from anything else in this article.

A research peptide is a single defined sequence. BPC-157 is GEPPPGKPADDAGLV, the same fifteen residues in every vial, characterised by that sequence.

Collagen peptides are hydrolysed collagen: a protein broken into fragments by acid, alkali or enzyme treatment. The result is a mixture of many different fragments of varying lengths and sequences, characterised by an average molecular weight distribution rather than by a sequence, because there is no single sequence to state.

So "peptide" in the two phrases does different work. One names a molecule. The other names a size class of protein fragments.

That distinction has a consequence for how each can be verified. You can confirm a defined peptide by mass spectrometry against an expected mass. You cannot do that for a hydrolysate, because there is no single expected mass, which is why those products are specified by molecular weight ranges instead.

Where Everything Lands

Growth Hormone, and Why It Is in a Different Conversation

Human growth hormone gets asked about in the same breath as the peptides on this site and it belongs somewhere else entirely.

hGH is 191 amino acids. Nearly five times the FDA line. It is a protein, a biological product, and it has an internal structure of four helices held by two disulfide bonds that has to be correct for the molecule to work.

Which puts it firmly in the analytical tier where purity figures stop answering the question. A correctly-sequenced hGH that folded wrong is inactive, and a chromatography peak will not distinguish it from one that folded right.

The reason people group it with peptides is functional rather than chemical: the growth hormone secretagogues, sermorelin and ipamorelin and the rest, are short peptides that act on the pituitary to change hGH output. Those are peptides. The hormone they act on is not. That distinction runs through the whole secretagogue comparison, where the entire premise is stimulating an endogenous protein rather than supplying it.

It is a good illustration of how the category label misleads. Sermorelin at 29 residues and hGH at 191 sit at opposite ends of the classification, opposite sides of the regulatory line, and in the same conversation, because what they have in common is a pathway rather than a chemistry.

Why Any of This Matters to a Buyer

Three things, and the third is the one I care about.

It tells you which regulatory conversation a compound is in. The July 2026 compounding review covered peptides, and no protein above 40 residues was on that agenda, because compounding rules for biologics work differently. That is set out in the legality reference.

It tells you what analytical methods are appropriate. A defined peptide can be confirmed by mass spectrometry against an expected mass. A protein above 40 residues has folding and, if expressed rather than synthesised, glycosylation, neither of which a purity figure detects. A hydrolysate has no single expected mass at all.

And it tells you when a supplier's category label is doing no work. If a page says peptide and the compound has no amino acids, the word is describing a shelf. That is not dishonest, and it becomes a problem the moment anyone reasons from the label to the chemistry, which is what happens whenever someone applies peptide storage guidance to a small molecule or expects a dinucleotide to behave like a chain of residues.

What This Article Does Not Settle

Whether 40 is the right number. The FDA chose it for administrative clarity rather than as a claim about where protein behaviour begins, and it acknowledged that proteins can be smaller.

Whether any molecule has been designed with the threshold in mind. Tirzepatide at 39 residues is one under it, which I have noted and am not going to build an argument on.

How other jurisdictions draw the line, which I have not researched. Everything here is United States regulation and should not be assumed to transfer.

What is well established: the FDA definition and its effective date, the reasoning the agency gave, that chemically synthesised polypeptides above 40 residues are biological products, and that insulin transitioned on that date. Those are checkable in the rulemaking documents.

Frequently Asked Questions

Is tirzepatide a peptide?

Yes, and only just by the regulatory definition. It is 39 amino acids, one under the FDA's 40-residue line, which makes it a drug rather than a biological product.

Where does a peptide become a protein?

Biochemistry has no hard line. The FDA does: greater than 40 amino acids is a protein, effective 23 March 2020. Forty or fewer is a peptide regulated as a drug.

Is Ozempic a peptide?

Semaglutide is 31 amino acids, so yes. It carries a non-natural residue at position 8 and a fatty diacid on lysine 26, neither of which changes the classification.

Is insulin a peptide or a protein?

A protein at 51 residues. It was deemed licensed as a biological product on 23 March 2020, without the molecule changing at all.

Are peptides steroids?

No. Steroids are built on a four-ring carbon skeleton from cholesterol and contain no amino acids. They also act on intracellular receptors rather than surface ones.

Is NAD+ a peptide?

No. It is a dinucleotide of about 663 daltons with no amino acids. It appears in peptide catalogues because those are organised commercially.

Are collagen peptides the same kind of thing?

No. They are hydrolysed collagen, a mixture of fragments characterised by average molecular weight rather than a single defined sequence.

Why does the FDA definition say "specific, defined sequence"?

Because it excludes mixtures. A hydrolysate has no single defined sequence, so it fails that clause whatever the fragment lengths are. The word "alpha" does similar work, leaving out backbone-modified chains such as beta-peptides and peptoids.

What changes analytically as peptides get longer?

Under about fifteen residues you can sequence directly by tandem mass spectrometry. Between fifteen and forty the problem is deletion sequences that co-elute with the correct product. Above forty, folding matters and no purity assay detects it.

Why does the classification matter?

It decides which statute applies. Peptides are drugs with generics; proteins are biological products with biosimilars. The FDA called the 40-residue rule a bright line chosen for clarity.

References