The objectives of this study were 1) to assess the validity of different databases which identify neural tube defect (NTD) cases in the population, and 2) to examine the temporal trends in NTD rates and the impact of prenatal diagnoses among pregnancies referred to a tertiary care hospital in Quebec City, Canada, from 1993 to 2002. Infant death and stillbirth certificates were a highly reliable source for ascertaining NTD cases, but their overall sensitivity was poor (13 percent). Med-Echo had very good sensitivity (92 percent), but there were many coding errors in the database and some diagnostic categories were not specific for NTD. The average NTD prevalence proportion was 6.5/1,000 births during the entire study period, decreasing from 12.2/1,000 in 1993 to 3.9/1,000 in 2002. Overall, 78.6 percent of NTD cases were diagnosed prenatally and the pregnancy was terminated in 52.6 percent of these. These two proportions were stable over the study years. To conclude, the combination of hospital discharge summaries and infant death and stillbirth certificates is a highly sensitive method for the ascertainment of NTD cases, including terminations of pregnancies, but medical records must be reviewed to exclude coding errors and to clarify unspecific diagnostic categories.