Congenital heart disease (CHD) is the most common fatal congenital anomaly in the first year after birth. Fetal echocardiography has become accepted as a method of diagnosing CHD in utero, but the benefits of prenatal diagnosis have not been evaluated. We performed a cohort analysis of neonates with CHD in the absence of other life-threatening conditions. One group (I) consisted of neonates who had been diagnosed as fetuses to have CHD and who were delivered at Yale-New Haven Hospital. The other group (II) consisted of neonates who were not diagnosed to have CHD until after birth, and were either delivered at, or transferred to, the same hospital. Costs of initial hospitalization, length of hospitalization, and survival to discharge home with the parents were primary outcome measures. Our primary hypothesis was that all outcome measures would be improved by prenatal diagnosis. We also examined secondary hypotheses that these same measures would be improved among selected subgroups, including those requiring univentricular management, those amenable to biventricular repair, those with ductal dependent lesions, and those requiring any surgery during the primary admission. From January 1, 1991 to June 30, 1996 we identified 45 antenatal cases and 54 postnatal cases of CHD that met the study entry criteria. The median length (+/- SE) of initial hospitalization was 16 +/- 3.8 days in group I, and 11 +/- 3.8 days in group II (p < 0.08). The median cost of initial hospitalization was $57,678 +/- 12,340 vs. $53,604 +/- 7249 (not significant). Eighty percent of group I survived to hospital discharge compared with 67% of group II (p = 0.14). There was no difference in survival among those requiring univentricular management (64 vs. 44%), but costs and length of hospitalization were greater in group I, regardless of whether or not surgery was performed postnatally. Surgery was undertaken more often among the prenatal diagnosis group with univentricular hearts (86 vs. 56%, p < 0.05). Among those fetuses amenable to biventricular repair, survival was better (96 vs. 76%, p < 0.05) and the cost of hospitalization was lower ($30,277 +/- 16,869 vs. $64,616 +/- 9441, p = 0.06) in the prenatal diagnosis group. Prenatal diagnosis of CHD does not result in the expected savings in cost, length of hospitalization or survival in the overall group of patients. The prenatal diagnosis group showed a skew towards single ventricle physiology. Within that group a high rate of pregnancy termination results in a group of parents who are likely to choose postnatal surgery despite high short- and long-term mortality risks.