PURPOSE To examine differences in correlates of neonatal and postneonatal infant mortality rates, across counties, by degree of rurality. METHODS Neonatal and postneonatal mortality rates were calculated from the 1998 to 2002 Compressed Mortality Files from the National Center for Health Statistics. Bivariate analyses assessed the relationship between neonatal and postneonatal mortality by Urban Influence (UI) codes. Multivariable, weighted least-squares regression models included measures of county socioeconomic conditions, health services and environmental risks. FINDINGS The bivariate analysis indicated neonatal and postneonatal mortality was significantly higher in the most nonmetropolitan counties compared to the most metropolitan counties. However the relationship was not linear across the Urban Influence codes. In the multivariable models, a nonmetropolitan advantage was observed for counties not adjacent to metropolitan areas for neonatal mortality. However, postneonatal mortality rates were higher in the most rural nonmetropolitan counties. CONCLUSIONS Certain characteristics of nonmetropolitan counties not adjacent to metropolitan counties and with an urban area of 2,500 population or more are protective against neonatal mortality (UI = 7, UI = 8). This may indicate that just having access to health services is more important to creating a protective effect for these nonmetropolitan counties than having a high concentration of medical facilities. The nonmetropolitan, not adjacent (UI = 9) disadvantage observed for postneonatal mortality supports the idea that the isolation of these areas combined with the combination of risk factors across the most nonmetropolitan counties leads to poorer postneonatal health outcomes in these areas.