The mortality pattern and causes of death in surgical neonates were evaluated over the period of 1980 through 1990 in order to assess whether any deaths might have been prevented. In the study period 1,010 neonates were admitted, 104 of whom died (10%). These 104 neonates formed our study population. The Clinical Classification System (CCS) and the Therapeutic Intervention Scoring System (TISS) were used to assess the severity of illness. All 104 nonsurvivors were seriously ill (CCS 3 and 4; TISS greater than 20). The mortality pattern was classified in three groups. Group A (nonpreventable death) included 56 deaths, 48 of which were due to the underlying disease. The other 8 patients died of a nonpreventable complication of treatment. Eleven of the patients in group B (permissible death) had (multiple) congenital anomalies associated with chromosomal anomalies. Three of them were not treated and in the other 8 treatment was initially started but later withdrawn. In most of the other 14 patients further treatment was withdrawn because of serious postoperative complications or inoperable cardiac anomalies. Group C (preventable death) consisted of 23 patients. Six of 10 had irreversible brain damage due to prolonged hypoxemia. In the other 13 patients of group C, death was due to sepsis acquired in the postoperative period and treated inadequately. Staphylococcus epidermidis was the most frequent isolated pathogen (7/13).