OBJECTIVE To measure population prevalence and determine potential predictors of neural tube defects. METHODS Analysis of all births reported to a mandated collection of perinatal data, and terminations prior to 20 weeks' gestation that have been reported to a data collection of birth defects in Victoria from 1983 to 1997. Prevalence at birth and risk ratios of infant and maternal characteristics associated with neural tube defects were calculated. RESULTS Prevalence of spina bifida has remained steady for 15 years and was 8.8/10,000 in 1997. Anencephaly increased to 7.9/10,000 in 1997. After exclusion of pregnancy terminations, the 1997 birth prevalence was 4.5/10,000 for spina bifida and 2.4/10,000 for anencephaly. Neural tube defects are identified in 1 in 1600 fetuses, the risk being significantly higher for epileptic women (Adjusted Odds Ratio (AOR) = 3.70, 95% CI 2.25-6.07), multiple births (AOR = 4.56, 95% CI 3.46-6.02), teenage mothers (AOR = 1.47, 95% CI 1.09-2.00) compared with those aged 25-29, and women with three or more previous pregnancies (AOR = 1.40, 95% CI 1.10-1.78). The risk was lower for women of East Asian (AOR = 0.70, 95% CI 0.49-1.00) and Middle Eastern origin (AOR = 0.60, 95% CI 0.35-1.02) and these differences were approaching statistical significance. CONCLUSIONS Total prevalence of neural tube defects did not decline up to 1997. CONCLUSIONS It is unlikely that targeting 'at risk' groups identified in this study would make a difference to neural tube defect incidence. However, consideration could be given to identifying larger 'at risk' groups such as those with homocysteine metabolism defects.