Abstract Several in vitro methods using both human and animal tissues have been developed to study hepatic metabolism and placental transfer. The pressure to minimize animal studies has increased during the past few decades due to the public opinion and ethical considerations. However, these methods need further evaluation of their predictive power when applied in vivo. The aim of this work was to produce new information of the metabolism and transplacental passage of several anticonvulsants as well as to evaluate the usefulness of the placental perfusion method and several in vitro methods for analyzing metabolism in the prediction of clinical pharmacokinetics. Carbamazepine (CBZ) metabolism was studied using human and mouse liver microsomes, human hepatocytes, human liver slices and yeast cells expressing recombinant enzymes. All test systems predicted well the major metabolite carbamazepine-10,11-epoxide (CBZ-E). Also, minor metabolites were produced in slightly variable amounts in all systems except cells with recombinant enzymes. All human liver systems demonstrated that CYP3A4 is the principal CBZ metabolising enzyme. However, our results on CBZ-treated mice suggested that the metabolism of CBZ to CBZ-E is mainly mediated by CYP1A1 in C57/BL6 mice. Autoinduction of CBZ metabolism was seen in hepatocytes and in incubations using microsomes from CBZ-treated mice. Human liver and mouse liver microsomes metabolized oxcarbazepine (OCBZ) mainly to its active metabolite, 10-hydroxy-10,11-dihydro-carbamazepine (10-OH-CBZ). Also, 10,11-trans-dihydroxy-10,11-dihydro-carbamazepine (10,11-D) and an unknown metabolite were detected. Placental transfer of lamotrigine (LTG) and diazepam (DZP) was considerable in the human placental perfusion system, indicating marked fetal exposure in vivo. The OCBZ, 10-OH-CBZ and 10,11-D analyzed from maternal venous and cord blood also suggested significant fetal exposure. The placental perfusion system predicts well the transplacental passage of LTG and OCBZ and its major metabolite. However, in vivo cord blood concentrations of DZP are higher than maternal concentrations. Placental perfusion studies did not predict this. Still, even with its limitations, the human placental perfusion method provides information that can be used to evaluate the risk factors associated with drug use during pregnancy because understanding of specific transport characteristics is a good basis for rational risk assessment. In conclusion, all of the tested in vitro systems were useful in the prediction of at least some aspects of in vivo pharmacokinetics and metabolism, but validation and refinement are still essential, as is also the need to keep in mind the limitations characteristic of each in vitro method.