The evidence is small. In the one milk study of injected semaglutide, eight mothers gave samples and none contained detectable drug [1]. The worst-case estimate put the infant’s dose at 1.26% of the mother’s. The Ozempic injection label still says there are no data on semaglutide in human milk [7]. The Ozempic and Rybelsus tablet labels say breastfeeding is not recommended [8]. The reason is an additive in the tablet, not semaglutide itself.
This page covers lactation only. Exposure during pregnancy is a separate question with a larger literature, set out in what the pregnancy evidence shows. Why pregnancy becomes likelier on these drugs is in the pregnancy log entry.
What each label says
Labels are read in this order: the product, then section 8.2, then the data behind it.
- Ozempic injection. There are no data on semaglutide in human milk, on the breastfed infant, or on milk production [7]. Semaglutide was present in the milk of lactating rats. Rat milk levels were 3- to 12-fold lower than maternal plasma. The label says the clinical relevance of the rat data is not clear.
- Ozempic tablets and Rybelsus. A clinical lactation study found semaglutide below the lower limit of quantification in milk [8]. SNAC and its metabolites were present in milk. Infants may clear SNAC more slowly than adults. The label advises that breastfeeding is not recommended on these tablets.
- Zepbound (tirzepatide). In a single-dose lactation study, tirzepatide in milk was undetectable or low [9]. There are no data on effects in the breastfed infant or on milk production.
The injection labels for Ozempic and Zepbound use the same closing sentence. The benefits of breastfeeding are weighed against the mother’s need for the drug and any potential effect on the infant. The label does not settle that weighing either way. The tablet format, and how it differs from the pen, is covered in how oral semaglutide is taken.
The semaglutide milk study
A milk bank released samples from eight women taking injected semaglutide [1]. Samples were taken at 0, 12 and 24 hours after a dose. The assay’s limit of detection was 1.7 ng/mL. Semaglutide was not detected in any sample.
The authors then assumed milk held drug at the limit of quantification. That is a worst case, since nothing was found. On that assumption the maximum relative infant dose was 1.26%. The standard safety threshold they cite is 10%.
LactMed adds the details the abstract leaves out [5]. The mothers took 0.25 to 1 mg a week. Their infants were 4 to 23 months old and mixed fed. They received breastmilk for 3 to 9 weeks during treatment. Normal growth and development were reported by the mothers. The weekly doses sit at the start of the ladder described in the semaglutide starting dose.
The tirzepatide data
Eleven healthy lactating women took one 5 mg dose of tirzepatide [9]. Milk was sampled over 28 days. Drug was undetectable in 164 of 171 samples. The other seven samples held less than 0.02% of the dose in total. The last measurable level came 5 days after the dose.
One dose is not weekly use. LactMed says levels at steady state are probably higher than this study shows [6]. It also reports a preprint of five women taking 2.5 to 5 mg weekly. No sample exceeded the limit of quantification. None of the mothers reported an adverse effect in the infant attributable to the drug.
LactMed’s summary is conditional. Tirzepatide is not a reason to stop breastfeeding if a mother requires it [6]. It should be used with caution until more data exist. The caution applies especially to a newborn or preterm infant.
How small the evidence is
Three reviews count the same gap from different directions.
- A 2026 systematic review included 36 studies of use around pregnancy and lactation [2]. Lactation data were sparse. It found one pharmacokinetic study, which reported no detectable semaglutide in milk.
- A 2026 scoping review with consensus guidance included 34 articles [4]. One was about lactation. Its introduction describes breastfeeding on these drugs as contraindicated.
- A 2026 review of diabetes drugs in breastmilk covered 20 drugs [3]. Five had human milk transfer data. Semaglutide was the only GLP-1 among them.
That third review also graded the guidelines. Most recommendations, 78%, were against glucose-lowering drugs while breastfeeding [3]. Only 4.4% were based on clinical evidence. The authors describe published recommendations as largely inconsistent.
So the sources point in different directions. LactMed lists semaglutide and tirzepatide as alternatives to each other. A consensus paper calls breastfeeding contraindicated. The injection labels leave the weighing open. The human milk data behind all three total 24 women.
Where this meets pregnancy and fertility
A breastfeeding mother may also be planning the next pregnancy. The Ozempic label says to stop at least 2 months before a planned pregnancy [7]. The reason given is the long washout period. Fertility evidence is also thin, and is set out in whether GLP-1 drugs affect fertility.
Postpartum contraception has its own label rule. The Zepbound label says tirzepatide may reduce the efficacy of oral hormonal contraceptives [9]. It advises a non-oral method, or an added barrier method, for 4 weeks after starting. The same applies for 4 weeks after each dose increase. The study behind that rule is in tirzepatide and oral contraception.
How this site handles the question
This site reports what the labels and studies say. It does not advise starting, stopping or continuing a drug while breastfeeding. That decision belongs to the prescriber, and the infant’s pediatrician is part of it.
Compounded semaglutide and tirzepatide are not FDA-approved. They carry no label of their own, so no section 8.2 applies to them. Whether a seller publishes a lactation policy at all is tracked in the questions they leave open.