The disposition of zidovudine (AZT) was investigated in near-term (day 20) pregnant rats after intravenous bolus administration of AZT at 50 mg/kg. A compartmental pharmacokinetic model was developed to describe AZT concentrations in maternal plasma (1), placenta (2), fetus (3), amniotic fluid (4), and the maternal tissue compartment (5). Model equations were fitted simultaneously to all concentration data by NONLIN least-squares regression. The model that best described the AZT concentration data (F test, AIC, sum of weighted squared residuals) incorporated bidirectional transfer between maternal plasma reversible placenta, placenta reversible fetus, placenta reversible amniotic fluid, and maternal plasma reversible tissue compartment. Transfer rate constants (1/h) were as follows: k12, 0.58 +/- 0.41; k21, 47.64 +/- 46.61; k23, 67.50 +/- 42.03; k32, 13.09 +/- 8.80; k24, 0.62 +/- 0.03; k42, 0.32 +/- 0.06; k15, 5.75 +/- 7.00; k51, 4.12 +/- 1.01; and k10, 1.51 +/- 0.80. AZT rapidly distributed into tissue and placenta compartments. However, AZT accumulated more slowly into amniotic fluid. Intercompartmental distributional clearances suggest that the mechanism of maternal-placental, placental-fetal, and fetal-amniotic fluid transfer of AZT was by passive diffusion. This maternal-fetal model for AZT may offer a useful approach for describing the placental transfer kinetics of other antiviral nucleosides as well.