This experiment was carried out with the aims of comparing the embryotoxic potential of valproic acid (VPA) in rats and mice, better defining the malformation pattern in these species, and comparing the embryotoxic effects detectable in mid-pregnancy to those observed in fetuses at term. Pregnant CD:Crl rats were treated subcutaneously (s.c.) at 08:00, 16:00, and 00:00 on day 9 of gestation with 0, 150, or 300 mg/ kg VPA; pregnant NMRI mice were treated s.c. at 00:00 on day 7 of gestation, and at 08:00 and 16:00 on day 8 of gestation with 0,75, 150, or 300 mg/kg VPA. Groups of females were killed on day 9 (mice) or day 11 (rats) of pregnancy and their embryos were carefully examined under a dissecting microscope. The remaining females were killed 1 day before parturition and their fetuses were examined for external, visceral, and skeletal malformations. A very high frequency (84%) of malformed embryos was recorded in the group of mice treated with 300 mg/kg, including open brain folds (73%), somite defects (36%), and heart malformations (20%). The rat embryos were less sensitive: only 43% of them were malformed after treatment with 300 mg/ kg, however, the pattern of malformations was quite similar to that observed in mice. The treatment with 150 mg/kg produced about 32% malformed embryos in mice and only 8.5% in rats. More than 84% of mouse fetuses from mothers treated with the highest dose showed a severe form of exencephaly. The axial skeleton was also severely affected. The postimplantation loss reached 52%. Exencephaly and skeletal malformations were also recorded in mouse fetuses from mothers exposed to 150 mg/kg. The dose of 75 mg/kg was without effects. Exencephaly was not observed in rat fetuses at term. In this species the axial skeleton was the most severely affected region at 300 mg/kg, while the lowest dose produced only sporadic malformations. These results confirm that the mouse is the more sensitive species for the teratogenic effects of VPA. Furthermore, it has been shown that, in both species, the axial skeleton is a system which is very sensitive to the teratogenic effects of VPA. The observed alterations show a possible link between axial specification and VPA and suggest further studies of embryos exposed to VPA for the expression of genes controlling the identity of vertebral segments.