Semaglutide and Breastfeeding: What the Milk Studies Actually Found

The question comes up constantly and almost never gets a straight answer: if you are breastfeeding, does semaglutide reach your baby through your milk?

It has an unusually clear answer, because somebody measured it. In the two studies that sampled real human milk, neither semaglutide nor tirzepatide was detectable. That is a genuine finding and it is worth knowing. It is also routinely overstated into “GLP-1s are safe while breastfeeding,” which is not what the data show and not what the labels say.

Here is what was actually measured, the one formulation where the answer flips, and the concern that the milk studies do not address at all — which, according to the research centre that ran them, is the one that actually matters.

What the milk studies measured

Two separate groups have put breast milk from women on GLP-1s through a mass spectrometer and looked for the drug.

The semaglutide study, published in Nutrients in 2024, came out of the InfantRisk Human Milk Biorepository at Texas Tech. Eight breastfeeding women on weekly subcutaneous semaglutide gave milk samples at 0, 12 and 24 hours after a dose — two at 0.25 mg, four at 0.5 mg, two at 1 mg. The assay could detect semaglutide down to 1.7 ng/mL.

It found none. Not a trace in any sample, at any dose, at any timepoint. Because you cannot compute an infant dose from a number you never measured, the authors did the conservative thing and assumed the milk was sitting just under the detection threshold the whole time. Even on that deliberately pessimistic assumption, the projected relative infant dose came to 1.26% — against a conventional safety threshold of 10%.

The tirzepatide data sit in the Zepbound label itself. Eleven lactating women received a single 5 mg injection and gave milk samples across 28 days — 171 samples in total.

In the tirzepatide study the seven samples that did register came to less than 0.02% of the maternal dose in total, and nothing was measurable after day five. The label is blunt that this still leaves gaps: there are no data on effects on the breastfed infant, and none on milk production.

Why the molecule does not get through

This is not a lucky result. It follows from what semaglutide is.

Most drugs that end up in milk are small, fat-soluble molecules that slip across the mammary epithelium without much trouble. Semaglutide is neither. It is a peptide of roughly 4,113 daltons — enormous by drug standards, dozens of times the size of the small molecules that transfer easily. Molecules that large cross into milk poorly and in tiny quantities.

Then there is a second barrier, and it is arguably the stronger one. Peptides are chains of amino acids, and an infant’s digestive tract treats a chain of amino acids the way it treats any other protein: it breaks it apart. Even if a meaningful amount reached the milk, an infant drinking it would digest it rather than absorb it. This is the same reason you inject semaglutide instead of swallowing it.

Two independent barriers, both working in the same direction, is why the measured result came back empty — and why it is unlikely to be a fluke of small samples.

The exception: the tablet is not the injection

This is the part that gets lost, and it is the one place where the answer genuinely reverses.

A peptide that an infant’s gut would digest is also a peptide an adult’s gut would digest. To make an oral semaglutide work at all, manufacturers pair it with an absorption enhancer called salcaprozate sodium, or SNAC, which helps the drug survive the stomach and cross into the bloodstream.

SNAC is a small molecule. It behaves nothing like semaglutide. And per the prescribing information, SNAC and its metabolites are present in human milk — with the added concern that newborns clear it more slowly than adults, so it could accumulate.

The result is a split that surprises people:

• The semaglutide injection label reports no data on semaglutide in human milk and uses the standard language — weigh the benefits of breastfeeding against the mother’s clinical need and any risk to the infant. It does not tell you to stop breastfeeding.
• The semaglutide tablet label says plainly: breastfeeding is not recommended during treatment. Not because of the semaglutide — the manufacturer’s own lactation study found semaglutide below the limit of quantification in milk — but because of the delivery agent wrapped around it.

So the drug everyone is asking about is not the problem in the oral version. The ingredient that makes the pill possible is. If you are weighing the pill against the injection, this is a distinction that matters far more while nursing than at any other time.

What the labels actually say

Worth setting these side by side, because “no data” and “not recommended” are very different statements and get quoted interchangeably.

One oddity worth flagging: the patient-facing Medication Guide for tirzepatide still tells patients the drug “may pass into your breast milk,” while the clinical section of the same document reports it was undetectable in 164 of 171 samples. Both statements are in the same label. If a pharmacist or an app has told you one thing and your clinician another, that mismatch may be why.

The concern the milk studies do not address

Here is the part that gets skipped, and the reason “it is not in the milk” is not the end of the conversation.

The InfantRisk Center — the same group that ran the semaglutide milk study — is direct about it: the main concern during breastfeeding is not drug transfer. It is whether you are getting enough calories, fluids and nutrients to support both yourself and your baby.

Think about what these drugs actually do. They work by suppressing appetite, and they are effective at it. Meanwhile exclusive breastfeeding carries a real daily energy cost on top of everything else your body is doing. Those two things pull hard in opposite directions.

A medicine designed to make you eat substantially less, taken during the one period when your intake is feeding two people, is a legitimate clinical question even with zero drug in the milk. The risks live in supply, hydration and nutrient adequacy — not in exposure. And no milk assay can measure any of that.

This is also why the honest version of the answer is not a yes or a no. It is: the exposure question looks reassuring, the nutrition question is open, and the second one is the one your clinician should be weighing.

Compounded products are a separate question

One further gap, and it is directly relevant if you are getting a compounded product rather than a brand one.

Everything above rests on studies of FDA-approved semaglutide and tirzepatide. Compounded versions are not FDA-approved, are not reviewed for equivalence, and have no lactation data of their own. One participant in the semaglutide study did use a compounded product, which is a single data point and not a basis for anything.

The molecular-size argument should hold for any genuine semaglutide regardless of who made it. But potency, purity and what else is in the vial are not things that argument covers, and none of it has been measured in milk. Worth saying out loud if compounded is the route in front of you.

What to ask before you decide

If you are nursing and considering a GLP-1, or already on one and have just discovered you are pregnant or nursing, these are the questions that actually move the decision:

• Injection or tablet? Not interchangeable here. The tablet label advises against breastfeeding; the injection label does not.
• What is my actual intake going to be? This is the real issue. Ask for specifics on calories, protein, fluid and micronutrients — not a general reassurance.
• How will we know if my supply drops, and what happens then? Agree the plan before starting, not after.
• How old is my baby? A newborn feeding exclusively is a very different situation from a toddler on mostly solid food.
• Can this wait? Sometimes the cleanest answer is timing rather than risk. If weaning is a few months out, that may settle it.
• Is a lactation consultant involved? A prescriber manages the drug. An IBCLC manages supply. This decision touches both.

Related reading: GLP-1 side effects covers the labelled adverse-event profile, and the pill versus the injection covers the formulation difference that matters most here.