Objective: To determine whether improvement in neonatal and infant mortality rates is possible or likely.

Setting: Regional neonatal intensive care unit.

Methods: Experience during a decade (1982-1991) was evaluated. We determined postnatal age at death and birth weight-specific and gestational age-specific mortality rates. Neonatal deaths (deaths before discharge) were categorized as "possibly preventable" or "probably unpreventable."

Results: Deaths occurring after 28 days ("postponed" deaths) contributed 9% of the total for the decade, and 5% for those with extremely low birth weight (ELBW; < 1000 gm) during the last 6 years; 47% of all deaths and 65% of deaths of ELBW infants occurred within 24 hours of birth. Congenital malformations accounted for 7%, 54%, and 66% of deaths when birth weight was 500 to 1499 gm, 1500 to 2499 gm, and > or = 2500 gm, respectively. In infants with birth weight > or = 1000 gm, probably unpreventable deaths (predominantly from congenital malformations, but also including hydrops and inborn errors of metabolism) accounted for 61% of deaths. Of deaths of ELBW infants, extreme prematurity (500 to 750 gm) accounted for 58%; major malformations and pulmonary hypoplasia contributed an additional 9%.

Conclusion: During the decade, the gestational age at which there was a 50% survival rate fell from 26 weeks to 24 weeks and a marked increase in the survival rate occurred at birth weights < 1500 gm (VLBW) after the introduction of exogenous surfactant therapy. The number of possibly preventable deaths is now very small. For any substantial impact on mortality rates, it will be necessary to lower VLBW and ELBW rates.