In 1991 a cut-off weight of 1 000 g and/or 28 weeks’ gestation for neonatal intensive care unit (NICU) admission was decided on by attending neonatologists at a Priorities in Perinatal Care Conference. These recommendations were not based on published evidence. At the time there were few data on the outcome of babies born in the public sector who received NICU. Aim . The aim of this study was to describe the demographic data (mother and baby) and outcomes of babies admitted to a tertiary NICU. Methods . During 1992 - 1996 (1992 cohort) and 1999 - 2000 (1999 cohort) two cohorts of babies treated in the NICU at Tygerberg Hospital, Western Cape, South Africa, were studied. Demographic data were collected prospectively on all admissions with a birth weight of less than 1 501 g and a gestational age of less than 32 weeks. Outcome data were survival, days of ventilation and NICU stay. Results . There were 455 babies in the 1992 cohort and 272 in the 1999 cohort. The mothers’ mean income was R892 per month and was higher in the 1999 cohort.The 1999 cohort comprised significantly smaller babies, at a mean birth weight of 1 119 g v. 1 198 g. The mean gestational age in the 1999 cohort was lower (29.2 v. 30.3 weeks), but so was the mortality rate (21.6% v. 26.1%). The main differences between the survivors and non-survivors were in their birth weight and gestational age and the mean income of their mothers. The mean number of ventilation days needed by these infants was low at 8.5 days, with an average stay in the NICU of 13 days. Discussion . Babies admitted to an NICU have a good chance of survival at a low mean number of ventilation and NICU days. The increase in survival in the 1999 cohort, in spite of low income, is in keeping with international trends and underlines the good short-term outcome of these small babies.