Is It Safe to Take a GLP-1 While Breastfeeding? What the Evidence Says

The answer in one paragraph

No study is large enough to certify a GLP-1 as "safe" while breastfeeding, and no manufacturer claims one is. But the honest summary has shifted, and in a direction most coverage has not caught up with: both molecules have now been measured in human milk, and both came back essentially undetectable. Eight nursing mothers on subcutaneous semaglutide gave milk samples at 0, 12 and 24 hours after a dose, and the drug was not detected in a single one⁴. Eleven lactating adults given one 5 mg dose of tirzepatide were undetectable in 164 of 171 samples². The NIH's Drugs and Lactation Database now states outright that only injectable forms of semaglutide should be used during breastfeeding — which is a recommendation about formulation, not an instruction to wean³.

Why "no data" was the default for so long

Lactating women are the last population anyone studies. A 2026 systematic review of every first- and second-line type 2 diabetes drug found human plasma-to-breastmilk transfer data for only 5 of 20 medications — semaglutide among them — and, more damningly, found that while 78% of clinical guideline recommendations advised against antiglycemic agents during breastfeeding, only 4.4% of recommendations rested on clinical evidence at all⁷. That is the machinery that produced the blanket "don't nurse on it" advice you have probably read. It was never a finding. It was a default in the absence of measurement, and the measurements have started arriving.

What was actually measured

The semaglutide figure deserves unpacking, because it is a worst case built on an assumption rather than a measurement. The authors could not detect the drug at all, so they took the assay's limit of quantification, assumed milk contained that much, and computed the relative infant dose from there: 1.26%, against the 10% threshold clinicians conventionally treat as reassuring⁴. The true figure is lower than that. They simply could not say how much lower.

Why the pharmacology predicted this

The chemistry was always in a nursing mother's favor. Semaglutide is a large peptide, more than 99% bound to albumin¹, and large, highly protein-bound molecules cross into milk poorly. Whatever trace did reach milk would then have to survive an infant's gut, which peptides largely do not — LactMed notes semaglutide's maximum oral bioavailability is around 1% in adults, the same reason these drugs are injected rather than swallowed³. The tirzepatide record makes the identical argument: absorption by the infant is unlikely because the drug is probably partially destroyed in the infant's gastrointestinal tract and poorly absorbed orally⁵.

What is still genuinely missing

Three specific gaps, each stated the way its source states it. Infant drug levels have never been published — LactMed's semaglutide and tirzepatide records both record "relevant published information was not found" under that heading as of their 2026 revisions^3,5. Effects on milk supply have never been published either, under the same heading in both records^3,5. And the milk studies captured single or short-term exposure rather than months of steady dosing; LactMed explicitly warns that the tirzepatide study does not account for the accumulation that would occur in typical use, so real-world milk levels are probably higher than it indicates⁵. A 2026 systematic review of GLP-1 use across preconception, pregnancy and lactation reached the same verdict, describing the lactation evidence as sparse and resting on that single pharmacokinetic study⁶.

What exists on the infant side is soft. The eight semaglutide mothers fed breastmilk for three to nine weeks while treated, and all reported normal growth and development in their infants; five tirzepatide-treated mothers continued breastfeeding and none reported adverse effects^3,5. That is maternal report in thirteen babies. It is not nothing, and it is not a follow-up study.

What LactMed actually recommends

This is the part most often garbled. Neither record tells a mother to stop. The semaglutide entry says the injectable forms are the ones to use during breastfeeding, steering away only from oral formulations because of the SNAC absorption enhancer they contain^3,1. The tirzepatide entry says that if a mother requires tirzepatide, it is not a reason to discontinue breastfeeding — while adding that until more data become available it should be used with caution, especially while nursing a newborn or a preterm infant⁵. Caution and prohibition are different instructions, and the difference matters a great deal when you are deciding whether to wean.

For what the labels themselves say, read what the label actually says about GLP-1s and breastfeeding; for the wider picture, see postpartum weight loss and GLP-1 timing. Whatever you decide, decide it with the clinician who knows your history and your baby's.

Where this sits on the evidence scale

Moderate for exposure, low for outcomes — two different questions that routinely get collapsed into one. That very little drug reaches milk is now measured rather than assumed, in two independent studies using validated assays, and it agrees with what protein binding and oral bioavailability predicted. That is roughly as good as pharmacokinetic evidence gets at this sample size. What remains genuinely unknown is whether any of it matters to a baby over months, because the infant data amount to thirteen mothers' impressions across a few weeks, no infant blood levels have ever been published, and nobody has measured milk during sustained steady-state dosing. A prospective cohort following breastfed infants of treated mothers for growth and development would settle it. Until someone runs one, "probably very low exposure, unproven infant safety" is the most a careful reader should take from this.