Abstract Study question Is the risk of stillbirth and neonatal death increased in pregnancies where clomiphene citrate was dispensed 3 months before or after time of conception? Summary answer The risk of stillbirth and/or neonatal death is 45% higher in pregnancies among women dispensed clomiphene citrate compared to unexposed women independent of confounders. What is known already Clomiphene citrate (CC) has been used for ovulation induction since 1967, despite prior evidence of fetotoxicity in humans and animals. CC has since been linked to fetal growth restriction, preterm birth, & congenital anomalies, which are independent risk factors for perinatal death. However, few studies have been large enough to provide data on infrequent adverse events. In high-income countries, maternal obesity is the greatest risk factor for perinatal death followed by advanced maternal age. These factors also contribute to infertility, but can be measured reliably and incorporated into safety studies to estimate patient and treatment effects. Study design, size, duration This is a whole of population cohort study involving 147,524 pregnant women and 242,079 births occurring from July 2003 to December 2015 in South Australia, assembled by linking records from the state-wide perinatal registry with Commonwealth government data on pharmaceutical dispensing. All post-20 week pregnancies have mandatory reporting. Clomiphene citrate dispensing by pharmacists resides in a single national database. Participants/materials, setting, methods Using de-identified pregnancy outcome data in South Australia, the prevalence of stillbirth (>20 weeks gestation) and neonatal death were examined in relation to clomiphene citrate exposure. Women were considered exposed if a prescription for clomiphene citrate was dispensed from three months before the conception window to end of first trimester. Dates for the conception window were derived from gestational age at birth (weeks), in turn based on date of last menstrual period and ultrasound scan. Main results and the role of chance There were 3,458 singleton pregnancies where women were exposed to clomiphene citrate, with 41 stillbirths and neonatal deaths recorded, a prevalence of 11.9 per 1000. For non-exposed pregnancies, the corresponding prevalence was 8.2 per 1000. In singleton pregnancies among women dispensed clomiphene citrate, compared to those without this exposure, stillbirth and neonatal death were increased by 45% (odds ratio (OR) = 1.45, 95% confidence interval (CI) 1.06 – 1.98). This association was strengthened after adjustment for maternal characteristics that were potential confounders: age, socio-economic status (based on postcode of residence), Aboriginal ethnicity, born outside Australia in low or middle income country, parity, smoking, pre-existing diabetes or hypertension, pregnancy-induced hypertension and pre-eclampsia (OR = 1.59, 95% CI 1.16 – 2.18). Obesity data was available from 2007 onwards and a restricted analysis including this variable saw an increased effect (OR = 1.79, 95% CI 1.19 – 2.72). It is implausible that a further unmeasured patient related factor could reverse this effect. The observed effect is also conservative as there were 200 multiple pregnancies in which women were exposed to clomiphene citrate, which were excluded as the number of deaths in offspring was too infrequent for robust stratified statistical comparisons. Limitations, reasons for caution Data reflect pharmacy dispensing, not consumption, although women are highly motivated to take this drug and over 5% of pregnancies were multiples, consistent with extensive use. Subfertility itself increases the risk of perinatal death and further analyses are underway to distinguish effects of the drug from indication for use. Wider implications of the findings Clomiphene citrate may not be as safe as presumed. This may partly be a consequence of variability in clinical prescribing and supervision, or the known variability in patient response to the drug. Our result is conservative as we excluded twins. It is supported by animal data and basic pharmacology. Trial registration number Not applicable