The teratogenicity of all-trans retinoic acid (RA) has been known for decades, however to date no extensive investigation of the dose-response and kinetics across a wide range of doses has been undertaken. The objectives of this study were to 1) increase the robustness of the dose-response curve for forelimb and cleft palate malformation resulting from RA exposure in gestation day (GD) 11 mice, 2) compare two techniques (computerized image analysis and visual morphological evaluation) for the assessment of fetal forelimb malformations, and 3) develop a physiologically based pharmacokinetic (PBPK) model to describe the maternal and fetal disposition of RA. Pregnant CD-1 mice were administered a single oral dose of all-trans RA (0, 2.5, 10, 30, 60 or 100 mg/kg) on GD 11. GD 18 fetuses were examined for malformations using visual morphological scoring and computerized image analysis. In separate experiments maternal plasma and fetal tissue time-course data for RA and a major metabolite (13-cis retinoic acid, 13cisRA) were collected on GD 11 and analyzed by HPLC. A PBPK model was then developed to describe the maternal and fetal disposition of RA and 13cisRA. Overall, computerized image analysis was not more sensitive in detecting changes in the humerus, radius and ulna than gross visual examination. Dose-response statistical analysis of developmental endpoints yielded comparable benchmark dose levels (5%) for long bones and cleft palate that ranged from 0.24 to 7.6 mg/kg all-trans RA. In general, the PBPK model predicted maternal plasma and embryonic concentrations of RA and 13cisRA compared favorably with experimental observations across the range of doses tested. Model predicted embryo concentrations of RA at the lowest two doses (2.5 and 10 mg/kg) were lower than the measured amounts and the model slightly over-predicted the clearance. This suggests that the kinetics of low doses of RA are somewhat different from the kinetics of the high doses administered in the study.