Objectives. To obtain population-based, clinical information regarding potentially modifiable factors contributing to death during the postneonatal period (28 to 364 days), we examined all postneonatal infant deaths in four areas of the United States to determine: (1) the cause of death from clinical and autopsy data rather than vital statistics, (2) whether death occurred during initial hospitalization or after discharge, and (3) the portion of postneonatal mortality attributable to infants who left the hospital with identified high-risk medical conditions. Design and Setting. Retrospective medical record review of all postneonatal infant deaths with birth weights greater than 500 g (total N = 386) born to mothers residing in: (1) the city of Boston (1984 and 1985, N = 55), (2) the city of St Louis and contiguous areas (1985 and 1986, N = 123), (3) San Diego County (1985, N = 112), and (4) the state of Maine (1984 and 1985, N = 96). Deaths were identified using linked birth and death vital statistics, and medical record audits of infants' and mothers' charts were performed. Causes of death were obtained from medical record review in conjunction with autopsy if performed (72%, N = 278), medical record alone (17%, N = 67), or vital statistics if no other source was available (11%, N = 41). The medical conditions at the time of discharge for each infant were reviewed and, if judged to confer an increased risk of morbidity or mortality, were classified as high risk. Results. The causes of death were sudden infant death syndrome (47%, N = 181), congenital conditions (20%, N = 77), prematurity-related conditions (11%, N = 43), infections (9%, N = 34), external causes (including injuries, drownings, ingestions, and burns) (7%, N = 25), and other (6%, N = 23). In 24% of congenital and 25% to 44% of prematurity-related deaths, infection was the acute or associated cause of death. Infants born to black mothers were more likely than those born to white mothers to die during the postneonatal period of all major causes of death (7.3 per 1000 vs 3.0 per 1000). Overall, 18% (N = 68) of deaths occurred to infants who never left the hospital; 79% (N = 305) of the infants were discharged before death; and discharge status was unknown in 3% (N = 13). Eighty-one percent of all infants with prematurity- related postneonatal deaths were never discharged, and of the total infants who were initially discharged, only 1% (N = 4) subsequently died of prematurity-related causes. Of all postneonatal deaths, only 16% (N = 62) left the hospital with identified high-risk medical conditions. Conclusions. These findings suggest that the etiology of postneonatal mortality is heterogeneous, with significant complexity in attributing specific causes of death and making designations of 'preventability.' The vast majority of infants who died of prematurity-related postneonatal causes never left the hospital, and only a small percentage of all infants that left the hospital before death were identified as being at high medical risk. Therefore, strategies for further decreasing postneonatal mortality must link high-risk follow-up programs to more comprehensive strategies that address risk throughout pregnancy and early childhood.