The infant mortality rate has been considered a summary of the quality of life and level of development of a given population. However, this indicator is very sensitive to such simple measures as oral rehydration therapy, vaccination, and continuation of breast-feeding. Given that such health activities have become more widespread, an infant mortality rate may no longer reflect a particular development model. With the aim of broadening the discussion regarding infant mortality, this study analyzed the 153 neighborhoods of the city of Rio de Janeiro, Brazil. Our objective was to identify areas with more risk, and the priority groups for interventions to decrease infant mortality. We analyzed neonatal and postneonatal mortality in each neighborhood. We also identified the children's home neighborhood and the location of their deaths and related these results to the socioeconomic classification of the corresponding neighborhoods. In relation to the average infant mortality rate for the city, we could not make statistically significant comparisons for some neighborhoods due to their small number of births. One-third of the infant deaths could have been prevented with early diagnosis and treatment. Only 15% of the deaths were considered unavoidable. Both neonatal mortality and postneonatal mortality were geographically dispersed, with no direct association with the socioeconomic profile of the neighborhoods. An analysis of the children's place of residence and the location of their deaths showed flows of patients from poor areas to more affluent city areas with better health services. This pattern highlights the effect of access to quality medical care on infant mortality.