This is the first in a series of studies investigating the developmental stage-specific neurobehavioral effects of all-trans retinoic acid (RA) exposure. Because high doses of this compound are known to be lethal to the developing organism, we first conducted a dose-response study to identify RA doses that produce low enough levels of gestational/postnatal mortality to make a behavioral analysis possible in survivors. Secondarily, at doses found to produce sufficient survivors on PND 28, effects on body and regional brain weights were examined. Finally, at these doses, effects on somatic malformations were evaluated. Four separate exposure periods were analyzed: gestational days (GD) 8 through 10, 11 through 13, 14 through 16, or postnatal days (PND) 3 through 5. In the postnatal exposure period rat pups were injected (s.c.) with three consecutive daily doses of 0, 5, 10, or 20 mg/kg RA on PND 3 through 5. This postnatal exposure had no detectable effect on survival, body or brain weight. In contrast, there was a marked sensitivity to RA in the GD 11-13 group. Many pups from dams given 10 mg/kg RA PO on GD 11-13 were found dead in the cage on the day of birth, and all surviving pups died within 4 days of birth. Examination of milkbands revealed no evidence of effective suckling in these short-term survivors. The same 10 mg/kg dose at GD 8-10 or GD 14-16 produced much lower mortality and pups appeared to suckle normally. To produce adequate PND 28 survival in the GD 11-13 group, it was necessary to reduce dosage to 2.5 mg/kg daily. Even this lower exposure produced effects on PND 28 body and brain weight, significantly lowering weights of body (84% of control), whole brain (94%), and cerebellum (90%). Cerebellar weight was also depressed as percent of whole brain weight, suggesting an effect focused specifically on this region. RA at 10 or 12.5 mg/kg over GD 14-16 also reduced cerebellar weight (92% and 91% of control, respectively). Thus, exposure on GD 14-16 had effects similar to those seen at GD 11-13, but only at considerably higher doses. In contrast, exposure to RA on GD 8-10 did not affect whole body or brain weight, and of eight brain regions examined, only brain stem weight was reduced (91% of control). The GD 8-10 exposure also differed substantially from later exposures in that it was the only treatment to produce substantial malformations, including exencephaly, eye and skeletal defects. We conclude that gestational exposure to RA produces lethality and regional brain stunting that is dose and developmental stage specific, with a pronounced sensitive period on GD 11-13. In contrast, the GD 8-10 period is most sensitive for production of malformations, albeit at somewhat higher doses.