Most questions about these drugs have one answer. This one has two, and which one a person is given depends on the indication printed on the box rather than on the drug in the pen.
Six FDA labels, read on 14 September 2026: Wegovy (the manufacturer's label effective 2026-06-18, re-read from DailyMed on 23 September 2026), Zepbound (2026-08-28), Saxenda (2026-02-25), Ozempic injection (2026-06-01, Novo Nordisk's copy, re-checked 2026-09-24), Mounjaro (2026-08-27, Eli Lilly's copy, re-checked 2026-09-24) and the shared Rybelsus and Ozempic tablets label (2026-01-30). Three obesity products, three diabetes products, two molecules between them.
Two brands, one molecule, two instructions
| Label | Indication | What section 8.1 instructs |
|---|---|---|
| Wegovy | obesity (also MASH) | discontinue when a pregnancy is recognised, for patients using it for weight reduction; for MASH, use only if the potential benefit justifies the potential risk |
| Zepbound | obesity | discontinue when a pregnancy is recognised |
| Saxenda | obesity | discontinue when a pregnancy is recognised |
| Ozempic | type 2 diabetes | use only if the potential benefit justifies the potential risk |
| Mounjaro | type 2 diabetes | use only if the potential benefit justifies the potential risk |
| Rybelsus / Ozempic tablets | type 2 diabetes | use only if the potential benefit justifies the potential risk |
Semaglutide appears in both halves of that table. So does tirzepatide. A patient taking 2.4 mg of semaglutide weekly for weight has a label telling her prescriber to stop the drug; a patient taking 2 mg of the same molecule weekly for diabetes has a label telling her prescriber to weigh it up.
The sentence that only appears on one side
The reason is written into the obesity labels and absent from the diabetes ones, in words that barely change between manufacturers:
Weight loss offers no benefit to a pregnant patient and may cause fetal harm.
Wegovy, Zepbound and Saxenda all carry a version of it. None of the three diabetes labels does. All three obesity labels then add the positive form of the same point: appropriate weight gain based on pre-pregnancy weight is recommended for all pregnant patients, including those with obesity or overweight, because of the obligatory weight gain that occurs in maternal tissues.
What the diabetes labels carry in that space is the other side of the scale. Poorly controlled diabetes in pregnancy, they record, raises the maternal risk of diabetic ketoacidosis, pre-eclampsia, spontaneous abortion, preterm delivery and delivery complications, and the fetal risk of major birth defects, stillbirth and macrosomia-related morbidity. They quantify it: a background risk of major birth defects of 6% to 10% in women with pre-gestational diabetes and an HbA1c above 7%, rising to 20% to 25% above an HbA1c of 10%, against 2% to 4% in the US general population.
So the divergence is not inconsistency. It is a risk-benefit judgement with one side empty in obesity and both sides occupied in diabetes, and the labels show their working. It is also the third time this year that reading two labels for the same molecule side by side has produced the finding — the same exercise on Wegovy and Ozempic showed one molecule carrying two different indication sets, and on Mounjaro and Ozempic it showed two cardiovascular indications resting on different kinds of trial.
Semaglutide says leave two months. Tirzepatide says the contraception may fail.
For anyone planning a pregnancy rather than discovering one, the two molecules give different information, and neither gives the other's.
Every semaglutide label instructs discontinuation at least two months before a planned pregnancy, and states the reason: the long washout period of semaglutide. Wegovy, Ozempic injection and the tablets label all carry it, in the prescriber sections and again in plainer words in the patient information.
Neither tirzepatide label contains any pre-pregnancy interval. There is no two-month rule, no one-month rule, nothing. What occupies the equivalent section of the Zepbound and Mounjaro labels is a different warning entirely: tirzepatide may reduce the efficacy of oral hormonal contraceptives through delayed gastric emptying, and patients using them are advised to switch to a non-oral method or add a barrier method for four weeks after initiation and for four weeks after each dose escalation. The pharmacokinetic figures behind that warning, and the 30-day version Foundayo's label carries, are set out with the other label interactions on GLP-1 drug interactions.
Set those two side by side and the failure modes are opposite. Semaglutide's labelled risk is that the drug lingers after it is stopped. Tirzepatide's is that a pregnancy may begin that was not planned at all.
A reader who switches between these drugs — and switching is common — will not be told the other one's warning, because each label describes only its own molecule. That is how labels are supposed to work. It is also why the two facts are worth printing on one page.
The animal data all six labels rest on
Human data in every one of these documents is described the same way: insufficient to evaluate a drug-related risk of major birth defects, miscarriage, or adverse maternal or fetal outcomes. The warnings therefore rest on reproductive toxicology, and in all three molecules the findings occurred at or below clinical exposure.
- Semaglutide. In pregnant rats dosed through organogenesis: embryofetal mortality, structural abnormalities and altered growth, at clinically relevant maternal exposures at the maximum recommended human dose. In rabbits and cynomolgus monkeys: early pregnancy losses and structural abnormalities, at clinically relevant exposures. The current label measures every multiple against the 7.2 mg weekly maximum; an older copy measured against 2.4 mg and so printed larger multiples for the same studies (two-fold or more in monkeys, for example).
- Tirzepatide. In pregnant rats: increased external, visceral and skeletal malformations, increased developmental variations and decreased fetal weights, at 0.5 times the human dose based on AUC. In rabbits: fetal growth reductions at clinically relevant exposures.
- Liraglutide. Teratogenic in rats at or above 0.8 times the human exposure; reduced growth and increased major abnormalities in rabbits at exposures below it.
Each label attaches the same caveat, and it is a real limitation rather than a formality: the effects coincided with pharmacologically driven reductions in maternal body weight and food consumption. In an animal given a drug whose purpose is to reduce food intake, an effect of the drug and an effect of the resulting maternal undernutrition cannot be separated by the study design. None of the labels claims to have separated them.
Registries
Two of the six labels name a pregnancy exposure registry. Zepbound's is run by Lilly and gives a phone number, an email address and a web address. Wegovy's gives a Novo Nordisk number and a website.
The other four — Ozempic, Mounjaro, Rybelsus/Ozempic tablets, and Saxenda — name none.
The Wegovy label now does one more thing the table above has to show: since it gained an indication for MASH, a liver disease, it carries both instructions in one document — discontinue for patients using it for weight reduction, weigh benefit against risk for MASH, because untreated advanced liver fibrosis may carry its own risk to mother and fetus. The rule this page describes, that the indication decides the instruction, now operates inside a single label.
What is not known
Almost everything a person in this position actually wants to know. Whether exposure in the weeks before a pregnancy was recognised carries any risk at all. Whether the animal malformations mean anything in humans. What happens to a pregnancy conceived one month after stopping rather than two. How many pregnancies have occurred on these drugs, and how they went.
The labels do not answer any of it, and they say so, in the same sentence in all six documents. Lactation is a separate section of the same labels, with its own small set of milk measurements and a split between the tablets and the injections; it is set out on GLP-1 drugs and breastfeeding. On a question asked this often, an absence that consistent is the finding — and it is the reason the two registries that do exist are worth more than any of the rodent data above.
This page reports what these documents record and is not medical advice; pregnancy decisions on any of these drugs belong with a prescriber, as our editorial standards set out.
Sources
FDA prescribing information published through openFDA, all read 14 September 2026 — section 8.1 Pregnancy, section 8.3 Females and Males of Reproductive Potential, section 7 Drug Interactions and section 13.1 in each: Wegovy (Novo Nordisk's label effective 2026-06-18, DailyMed set ee06186f-2aa3-4990-a760-757579d8f77b, re-read 23 September 2026), Zepbound (2026-08-28), Saxenda (2026-02-25), Ozempic injection (2026-06-01, Novo Nordisk's copy, re-checked 2026-09-24), Mounjaro (2026-08-27, Eli Lilly's copy, re-checked 2026-09-24), Rybelsus and Ozempic tablets (2026-01-30).
- Corrected 2026-09-23: the Wegovy label first read for this page (effective 2024-04-23) was a repackager's copy returned by an openFDA brand-name search. Novo Nordisk's current label (effective 2026-06-18) states the registry in the present tense ("There is"), adds the MASH indication with its own pregnancy instruction, and measures the animal findings against the 7.2 mg maximum dose. The page is updated throughout; an earlier section built on the older copy's future-tense registry sentence is removed.