Term infants with seizures and evidence of perinatal asphyxia were prospectively identified in 1 city and 2 time periods: 1978-1981 and 1991. Infants with multiple congenital abnormalities, hypocalcaemia or infection were excluded. Although there was little change in the overall incidence of neonatal seizures between 1978-1981 (1.9 per 1,000) and 1991 (1.78 per 1,000, N.S.) there was a marked reduction in small for dates infants with seizures: 8 of 19 infants in 1978-1981 compared to none of 16 in 1991 (p < 0.005). In contrast, infants > or = 41 weeks continued to show a markedly increased risk for asphyxia (relative risk 4.48, 95% CI: 1.7-12.3). The mechanism of this improved outcome for small for gestational age infants is unknown, but speculatively may be due to improved obstetric monitoring techniques allowing early identification of compromised infants.