A detailed retrospective analysis was made of the records of 486 preterm infants, who accounted for 5-1% of all births during 1973 and 1974. Whereas preterm delivery did not contribute to perinatal mortality in terms of stillbirth, it outweighed all other causes in terms of early neonatal deaths. Preterm birth was responsible for 85% of the early neonatal deaths not due to lethal congenital deformities. Early neonatal mortality rates were closely linked both to gestational age and birth weight and to the reason for preterm birth. Early neonatal mortality was high (97 per 1000) when preterm labour was spontaneous, whether or not associated with material or fetal disease or with multiple pregnancy, but low (27 per 1000) when preterm delivery was elective. Preventing spontaneous preterm labour would considerably reduce neonatal mortality in our community.