Objective To investigate the transplacental transfer and the effects of protein binding on the transfer of citalopram, desmethylcitalopram, fluoxetine and desmethylfluoxetine in the isolated perfused human placenta model. Design Prospective observational study.Methods Fifteen term human placentas were obtained immediately after delivery with maternal consent and a 2-hour non-recirculating perfusion cycle of a single placental cotyledon was set up. Citalopram (1230 nmol/L) and desmethylcitalopram (600 nmol/L) or fluoxetine (1455 nmol/L) and desmethylfluoxetine (1525 nmol/L) were added to the maternal reservoir and their appearance to the fetal circulation was followed by repeated measurements. To investigate the effect of protein binding on the transfer of citalopram and fluoxetine, nine additional perfusions were performed without albumin in the perfusion medium. Citalopram and desmethylcitalopram concentrations were measured by reversed-phase high performance liquid chromatography. Fluoxetine and desmethylfluoxetine concentrations was measured by gas chromatography and antipyrine (used as a reference compound) concentrations spectrophotometrically. Results The mean (SD) steady-state transplacental transfer (TPT SS %) for citalopram, desmethylcitalopram, fluoxetine and desmethylfluoxetine was 9.1%, 5.6% ( P ¼ 0.017 compared with citalopram), 8.7% and 9.1%, respectively, calculated as the ratio between the steady-state concentrations in fetal venous and maternal arterial sides. The TPT SS %s of citalopram, desmethylcitalopram, fluoxetine and desmethylfluoxetine were 86%, 50%, 88% and 91% of that of freely diffusable antipyrine. The absence of albumin significantly reduced the transfer of citalopram and fluoxetine (TPT SS % 1.1% and 4.8%, respectively) but not the transfer of antipyrine. Conclusion Citalopram, fluoxetine and desmethylfluoxetine all cross the human placenta, and may, therefore, affect the perinatal outcome of infants exposed to these drugs during pregnancy. The transfer of desmethylcitalopram was significantly lower, which in the clinical setting may suggest lower fetal exposure of serotonin re-uptake inhibition by citalopram compared with fluoxetine. The presence of albumin was necessary for the transplacental transfer of both citalopram and fluoxetine.