INTRODUCTION: Presently, no well-validated predictive tools are available for human placental transfer. We studied the transplacental passage of diazepam (DZP) in a recirculating dual human placental perfusion and compared the data with in vivo clinical data from the literature. METHODS: Term placentas from healthy mothers without medication were used. The dual, recirculating perfusion technique was used. DZP (2 microg/ml, n = 4; 200 ng/ml, n = 3) and the reference compound antipyrine (100 microg/ml) were added into the maternal circulation simultaneously. The disappearance of drugs from the maternal circulation and appearance into the fetal circulation were followed every 15 min for 2 h. RESULTS: DZP was detectable in the fetal circulation within 15 min in all of the perfusions indicating rapid transfer. DZP concentrations in the maternal circulation were higher than in the fetal circulation throughout the perfusion with both initial concentrations. At the end of the perfusion, the feto-maternal ratio was 0.48 +/- 0.11 (mean +/- S.D.) and the transfer from the maternal to the fetal compartment 18.4 +/- 3.6% with 2 microg/ml of DZP and 0.55 +/- 0.10 and 20.5 +/- 3.1% with 200 ng/ml of DZP, respectively. DZP concentrations in the perfused area of the placenta were in average 2 times higher than in the maternal perfusate and 3.6 times higher than in the fetal perfusate. Total recovery of DZP from samples, perfusion fluid, and perfused tissue was 37.6 +/- 21%. DISCUSSION: Since animal studies in vivo do not accurately predict human placental transfer and it is problematic to study placental transfer of drugs in humans in vivo, the present human placental perfusion system could serve as one part of a test battery for fetotoxicity. However, although our earlier studies and those from the literature indicate a good correlation between in vivo and placental perfusion data, the present study shows this is not the case for all drugs.