Objective. To describe the epidemiological impact of prenatal diagnosis and selective abortion on the frequency of neural tube defects (NTD) in the period 1980-1992 in the Northern Netherlands in comparison with data from other European regions. Design. Descriptive. Setting. 17 'European registration of congenital anomalies' (EUROCAT) registrations, localized in 10 European countries. Method. Data were collected actively and retrospectively from multiple sources fed by voluntary registration of congenital anomalies in live births, stillbirths and pregnancies terminated because of congenital anomalies. Results. In Europe the total birth prevalence of NTD in the period 1980-1992 ranged from 5.3 per 10,000 in Switzerland to 29.0 per 10,000 in Glasgow, a difference of a factor 5.5. In live births the difference was ninefold: ranging from 2,0 per 10,000 in Paris to 18.8 per 10,000 in Dublin. The Netherlands had a conspicuously high prevalence among live births, higher than in other regions in continental Europe. For spina bifida the live birth prevalence both in other continental regions and in Glasgow was also lower than in the Netherlands. In Glasgow serum alpha-foetoprotein screening apparently led to frequent early prenatal diagnosis of NTD and to frequent termination of pregnancy. In Paris the use of ultrasound screening appears to lead to frequent later prenatal diagnosis, as well as frequent termination of pregnancy. Conclusion. In the Netherlands the impact of prenatal diagnosis and selective abortion is limited, so that primary prevention (periconceptional use of folic acid) is more important than in some other European countries.