The purpose of this study was to characterize the disposition of zidovudine in the maternal-fetal-amniotic fluid unit in vivo. Zidovudine was administered as a constant-rate infusion or a bolus dose to the dam, fetus and amniotic cavity (bolus dose only) of the near-term pigtailed macaque model. A fetal-maternal plasma steady-state concentration ratio of 0.76 +/- 0.06 suggested that the drug was transferred extensively to the fetal compartment. Similarly, the mean fetal-maternal plasma area under the curve (AUC) ratio after administration of an i.v. bolus dose to the dam was 0.84 +/- 0. 09. Both ratios were significantly less than unity (P < .05), which indicates that fetal exposure to zidovudine was lower than maternal exposure. The placental transfer of zidovudine was passive, with a clearance of approximately 2.0 ml/min/kg, about 35% of the rate of the placental blood flow marker antipyrine. Zidovudine concentration in the amniotic fluid was higher than that in the fetal plasma because the drug is eliminated slowly from the amniotic cavity. The steady-state and i.v. bolus experimental designs resulted in close estimates of the extent of placental transfer of zidovudine (steady-state fetal-maternal plasma concentration ratio or fetal-maternal plasma AUC ratio), which indicates that the extent of transfer of zidovudine is independent of the mode of drug administration. We predict that when zidovudine is administered orally to pregnant women, the average fetal exposure to zidovudine will be approximately three fourths of the maternal exposure. This observation suggests that the dose administered to the pregnant woman need not be changed even if the fetus is the primary target of therapy.