We present nationally representative estimates of neonatal seizure risk by gender, race and geographic region of the United States. National Hospital Discharge Survey data were analyzed for the period 1980-1991. Birth-weight-adjusted risks of neonatal seizures were calculated by the direct method for each gender or race group and for each census region by 4-year intervals. The overall risk of neonatal seizures was 2.84 per 1,000 live births. Risk estimates were consistently higher in low-birth-weight infants (relative risk 3.9). Unadjusted risks were similar across race and gender groups; birth weight adjustment had very little effect. No clear temporal trend was apparent over the 12-year study period. National Hospital Discharge Survey data provide reasonable, although conservative, estimates of neonatal seizure risks nationwide. Underascertainment of neonatal seizures, particularly among sick low-birth-weight infants, is likely due to data collection limitations of the National Hospital Discharge Survey.