Data from 95 heart transplantations performed at The Johns Hopkins Hospital from July 1983 to October 1988 were analyzed to detect patterns of morbidity and mortality. Using nonparametric techniques, hazard functions were determined for all deaths and for deaths due to infection or rejection. The rates of rejection and infection (episodes per patient-month) were determined within each of ten intervals following transplantation. A total of 19 deaths, 281 rejection episodes, and 180 distinct infections were available for analysis during a follow-up of 1 to 62 months. The hazard function for rejection appeared biphasic, with a rapidly decelerating early phase during the first year followed by a constant late phase. The hazard function for infection was triphasic, with a delayed, decelerating early phase, a period of increased risk approximately 2 years after operation, and finally a late constant phase. Both infection and rejection rates (episodes per patient-month) were biphasic, with rapidly decelerating early phases and constant late phases. Multiple regression analysis demonstrated that eventually nonsurviving patients had significantly higher rates of rejection and infection during both the early and late phases compared with survivors. The increased rate of rejection among nonsurvivors was evident throughout follow-up, although no deaths were attributable directly to rejection after the first 8 months. These data suggest that a complex interrelationship between infection and rejection determines late survival after cardiac transplantation and that aggressive treatment of late rejection predisposes toward death from infection.