Lactation is the point on the reproductive timeline where the labels in this class stop agreeing with each other. Two molecules, four widely dispensed products, and four different sentences about the same question. One says nothing is known. One says a study was done and nursing is still not advised. One says the drug was measured in milk and barely found. None of them says what happened to a baby.
That gap matters more here than almost anywhere else on this site, because the postpartum months are exactly when the demand peaks. The labeled rules that apply before conception and during pregnancy are a separate question with a separate evidence base, set out in the pregnancy and contraception article.
What the four labels say, read this week
Read on DailyMed on September 15, 2026, the Wegovy label splits by formulation. For the injection it states that there are no data on the presence of subcutaneously administered semaglutide or its metabolites in human milk, on the effects on the breastfed infant, or on the effects on milk production. The Ozempic label carries the same absence, adding only that in lactating rats semaglutide was detected in milk at levels 3- to 12-fold lower than in maternal plasma, with the clinical relevance of that comparison described as unclear because lactation physiology differs between species.
The tirzepatide labels are the outlier, and they are the outlier in the direction nobody expects. Both the Zepbound and Mounjaro labels report a single-dose clinical lactation study in which the concentration of tirzepatide in breast milk was found to be either undetectable or low compared with the maternal administered dose. Both add the same qualifier the semaglutide labels carry: there are no available data on the effects of tirzepatide on the breastfed infant or on milk production.
All four labels then hand the decision back, in identical language, by instructing that the developmental and health benefits of breastfeeding be weighed against the mother’s clinical need for the drug and any potential adverse effects on the breastfed infant. That is a framework, not an answer, and it is the same framework whether the underlying evidence is a measurement or a blank.
The two human measurements, and how small they are
The tirzepatide figure comes from the label’s own data section. Following a single subcutaneous 5 mg dose in 11 healthy lactating adult females, tirzepatide was undetectable — at a milk limit of detection of 4 ng/mL — in 164 of 171 samples assayed. The cumulative amount detected in the remaining seven samples across a 28-day sampling window was equivalent to less than 0.02% of the maternal administered dose, with the last measurable concentrations five days after the dose. The label states that an area-under-the-curve figure could not be calculated at all, because there were not enough quantifiable concentrations to compute one.
The semaglutide figure comes from a journal rather than a label. Milk samples from eight women, released from the InfantRisk Center human milk biorepository and collected at 0, 12 and 24 hours after semaglutide administration, were assayed by high-resolution liquid chromatography-mass spectrometry with a limit of detection of 1.7 ng/mL and a limit of quantification of 5.7 ng/mL. Semaglutide was not detected in any sample. Substituting the limit of quantification as a worst-case milk concentration produced a projected relative infant dose of 1.26%, against the conventional 10% threshold.[1]
Nineteen women. That is the entire published human transfer record for a drug class dispensed to millions of people, and each of those women contributed data around a single dose rather than at steady state on a maintenance regimen. A pooled review of GLP-1 and dual agonist exposure across preconception, pregnancy and lactation screened the literature to September 2025, included 36 studies in total, and described the lactation evidence in one clause: sparse, with a single pharmacokinetic study reporting no detectable semaglutide transfer into human milk.[2]
What a relative infant dose does and does not answer
A relative infant dose expresses the amount an infant receives through milk as a percentage of the mother’s weight-adjusted dose, and figures under 10% are conventionally treated as unlikely to matter. The 1.26% above is not a measured value. It is what the number would be if semaglutide were sitting just below the detection threshold in every sample, which is the most pessimistic assumption the data permit.[1]
The limit of that arithmetic is that it answers exposure and not effect. It says a nursing infant is very unlikely to receive a pharmacologically meaningful dose. It does not say what happens to a baby fed by someone losing weight rapidly, whether milk volume holds, or whether milk composition shifts under a sustained calorie deficit. Those are different measurements, and nobody has published them.
The formulation that was studied is the one advised against
The Wegovy label’s tablet section is the part of this evidence base that runs against the grain. It reports that a clinical lactation study with the semaglutide oral tablet formulation found semaglutide concentrations below the lower limit of quantification in human milk — the same reassuring result as the injection studies — and then advises patients that breastfeeding is not recommended during treatment.
The reason is not the peptide. It is SNAC, the absorption enhancer that makes an oral peptide viable at all, which the label states is present in human milk along with its metabolites. Because the enzymes that clear SNAC may be less active in infants than in adults, higher plasma levels could occur in neonates, and because semaglutide formulations without SNAC exist, the label routes around the question entirely. The general trade-offs of swallowing this molecule rather than injecting it are in the oral versus injectable comparison.
So the one semaglutide product with a published human lactation study carries the most restrictive lactation instruction in the class, and the product with no human data at all carries the permissive weigh-the-benefits language. More measurement produced a stricter rule, because what got measured was an excipient nobody had been worrying about.
The wider record is not just thin, it is inconsistent
A systematic review of plasma-to-breastmilk transfer across drugs used for type 2 diabetes found clinical transfer data for only 5 of 20 agents — metformin, glyburide, glipizide, tolbutamide and semaglutide. Of the guideline recommendations it assessed, just over half (51.7%) gave explicit guidance, and only 4.4% were based on clinical evidence, while 78% advised against use of antiglycemic agents while breastfeeding.[3]
That is the shape of the problem stated numerically. The overwhelming majority of published advice on this question is against the drug, and the overwhelming majority of that advice rests on something other than a measurement. Where an instruction has no data underneath it, the instruction may still be right — it is simply not evidence, and it should not be quoted as though it were.
What a prescriber is actually weighing
Three inputs, and only one of them has a number attached. The first is transfer, which the measurements above address and which looks small for both molecules. The second is infant effect, which no study has reported. The third is the maternal indication, which is where the real argument sits: postpartum weight retention is a genuine health problem with consequences for later pregnancies, and a European position statement on women with obesity across reproductive life names postpartum weight management as necessary while stating in the same document that clinical data on the safety and efficacy of obesity medication during pregnancy or lactation are limited.[4]
A fourth consideration belongs in the conversation and has no literature behind it at all. These drugs work by reducing energy intake, and lactation raises energy requirement. Nobody has run the study that would say whether that combination affects milk volume or composition, which makes it a question for the clinician who can monitor an individual infant’s growth rather than a question a checkout page can settle.
Timing is the other half. Stopping a drug is not a neutral act — weight returns after withdrawal — so a plan that defers treatment until weaning has a foreseeable cost attached, and a plan that treats during lactation rests on evidence that does not exist yet. Neither of those is a decision a seller is positioned to make. Sellers who market specifically to women are collected on their own board, and the medication itself is not a different one.
None of the quoted text travels with a compounded vial
Every sentence quoted above comes from an FDA-approved label for a branded product. Most cash-pay sellers dispense compounded semaglutide or tirzepatide, which is not FDA-approved and which the FDA does not review for safety, efficacy or quality before it is dispensed — see what a compounded vial contains. A compounded preparation may also carry excipients the branded product does not, and the SNAC example above is precisely the case where an excipient rather than the active drug drove the lactation instruction.
How thoroughly an intake form asks about nursing varies by seller, which is one of the things the provider write-ups record, against the criteria set out in the methodology.
The honest summary
What is established: both molecules appear in human milk at concentrations at or below the limits of the assays used to look for them, measured in 19 women across two single-dose studies, and the worst-case exposure arithmetic lands far under the threshold that normally prompts concern. What is not established: any effect on a breastfed infant, any effect on milk production or composition, anything about chronic dosing, and anything at all about liraglutide, dulaglutide or the compounded preparations most readers are actually buying.
No label in the class recommends starting one of these drugs while nursing, and none of them forbids it except the oral semaglutide tablet, for a reason that has nothing to do with the peptide. Between those two positions sits a decision that belongs to a prescriber who knows the infant, the feeding pattern and the maternal indication — not to a milk concentration, and not to a checkout.