Curly tail mouse mutant embryos (ctlct) develop spinal neural tube defects (NTD) in 54% of cases, comprising isolated tail flexion defects and spina bifida with tail flexion defects. Both types of spinal NTD result from delayed closure of the posterior neuropore (PNP). Previous studies (Seller et al. El9791 Proc. R. SOC. Lond. Biol. 20695-107; Seller and Perkins [19821 Prenat. Diagn. 2297300) described a paradoxial effect of retinoic acid (RA) on the phenotypic expression of the ct mutation: Treatment with low doses of RA on day 8 of gestation increased the incidence of total NTD, whereas low doses of RA administered on day 9 resulted in reduced incidence of total NTD. In order to investigate further the reported preventive effect of RA, we have carried out detailed analyses of the effects of maternal treatment with 5 mg/kg RA on the incidence of NTD at different developmental stages, and on the development and growth of ctlct embryos. We found that 5 mg/kg RA reduces the incidence of spinal NTD in a stage-specific manner, without increasing the incidence of cranial NTD. The effect of RA is specific: There were no other alterations in morphogenesis, growth, development, resorption rate, or litter size. RA was more effective in the prevention of isolated tail flexion defects than of spina bifida. Prevention of isolated tail flexion defects was maximal (50% reduction) when RA was administered between 10 days 4 hours and 10 days 8 hours post coitum (P.c.) inclusive (24 to 34 somite stage). In contrast, maximal prevention of spina bifida (36%) resulted from RA administration at 10 days 8 hours P.c.; decreased PNP size in treated embryos, compared with control embryos, was evident by 6 hours after the treatment at the 27 to 31 somite stage. The preventive effect of RA on spina bifida was related to maternal phenotype: By comparison with phenotypically straighttailed mothers, curly-tailed mothers had a greater incidence of spina bifida among their offspring, and this incidence was unresponsive to RA. This result suggests that additional factors, such as modifier genes, modulating phenotypic expression of the ct gene, may be present in the ct mutant stock. Here we show that the RA prevention of spinal NTD is a specific effect. Hence it is plausi-0 1994 WILEY-LISS, INC. ble that this prevention of spinal NTD by RA is mediated via nuclear retinoic acid receptors.