A collaborative effort to assess factors affecting newborn survival at neonatal intensive care units (NICUs) was made by studying 1948 newborns admitted to nine NICUs in the city of São Paulo between 1 June and 30 November 1991. Data on the study subjects were obtained using a standardized form. This was the first activity undertaken by a network of neonatologists (the Paulista Collaborative Group on Neonatal Care) dedicated to jointly evaluating and improving neonatal care in that city. The study results showed an overall mortality of 59 deaths per 1000 neonates, with survival improving as gestational age and birthweight rose. Other variables significantly affecting survival were a poor maternal obstetric history (a previous stillbirth or neonatal death, or two or more spontaneous abortions); birth asphyxia (Apgar at 5 minutes <7); respiratory distress syndrome; severe infections; and major malformations. However, multiple logistic regression analysis showed that the rates of neonatal survival in the nine NICUs differed even when these factors were considered. Potential sources of this variability included undetermined population differences in neonatal disease severity and medical care. These results suggest a need for greater efforts to identify and reduce risk factors associated with neonatal mortality, and to adequately evaluate the medical care provided in NICUs. Within this context, the collaborative network of neonatologists established in São Paulo provides a sound organizational structure for evaluating and improving the effectiveness of neonatal care.