AIMS: To describe trends in total and live birth prevalence, regional differences in prevalence, and outcome of pregnancy of selected congenital anomalies. METHODS: Population based registry study of 839,521 births to mothers resident in five geographical areas of Britain during 1991-99. Main outcome measures were: total and live birth prevalence; pregnancy outcome; proportion of stillbirths due to congenital anomalies; and secular trends. RESULTS: The sample consisted of 10,844 congenital anomalies, giving a total prevalence of 129 per 10,000 registered births (95% CI 127 to 132). Live birth prevalence was 82.2 per 10,000 births (95% CI 80.3 to 84.2) and declined significantly with time. The proportion of all stillbirths with a congenital anomaly was 10.5% (453 stillbirths). The proportion of pregnancies resulting in a termination increased from 27% (289 cases) in 1991 to 34.7% (384 cases) in 1999, whereas the proportion of live births declined from 68.2% (730 cases) to 58.5% (648 cases). Although similar rates of congenital anomaly groups were notified to the registers, variation in rates by register was present. There was a secular decline in the total prevalence of non-chromosomal and an increase in chromosomal anomalies. CONCLUSIONS: Regional variation exists in the prevalence of specific congenital anomalies. For some anomalies this can be partially explained by ascertainment variation. For others (neural tube defects, diaphragmatic hernia, gastroschisis), higher prevalence rates in the northern regions (Glasgow and Northern) were true differences. Live birth prevalence declined over the study due to an increase in terminations of pregnancy.