This study addresses two questions about why neighborhood contexts matter for individuals via a multilevel, spatial analysis of birthweight for 101,662 live births within 342 Chicago neighborhoods. First, what are the mechanisms through which neighborhood structural composition is related to health? The results show that mechanisms related to stress and adaptation (violent crime, reciprocal exchange and participation in local voluntary associations) are the most robust neighborhood-level predictors of birth weight. Second, are contextual influences on health limited to the immediate neighborhood or do they extend to a wider geographic context? The results show that contextual effects on birth weight extend to the social environment beyond the immediate neighborhood, even after adjusting for potentially confounding covariates. These findings suggest that the theoretical understanding and empirical estimation of 'neighborhood effects' on health are bolstered by collecting data on more causally proximate social processes and by taking into account spatial interdependencies among neighborhoods.