Objective To study the teratogenicity of valproic acid (vpa) in Wistar rats. Methods Twenty four pregnant Wistar rats were randomly assigned into 5 groups: 3 VPA of different dosages groups (400 mg/kg, 450 mg/kg and 500 mg/kg groups), the normal saline (NS) control group and the Blank control group. On the 9th day of gestation, 3 VPA groups were subcutaneously injected different dosages of vpa at 9am and 4pm. The NS control group was administrated 1 ml of NS instead and the Blank control group had no intervention. On the 20th day of gestation, the embryos were taken out and the specimens from spines were double stained with Alcian blue GX and Alizarin red S. The distance of the two cartilaginous ends of the vertebra arch was measured with a stereomicroscope and compared with the normal limit. Results In the Blank control group, the distance of the two vertebral cartilaginous ends was less than 166.4 μm (the superior limit of 112.0±1.96×27.7 μm) from the 9th thoracic to the 3rd sacral vertebra; less than 185.5 μm (the superior limit of 127.7±1.96×29.5 μm) in the 4th sacral vertebra; and less than 198.7 μm (the superior limit of (142.1±1.96×28.9 μm) in the 5th sacral vertebra. There was no malformation in the NS control group. The incidence of spina bifida occulta was 80%, 93% and 100% in the VPA 400 mg/kg, 450 mg/kg and 500 mg/kg groups respectively. The craniofacial, skull, spine and tail malformations occurred in a fetus in the vpa 500 mg/kg group. Conclusions Vpa could induce spina bifida occulta in foetal rats.