BACKGROUND: We study geographic variation within one community in the City of Calgary using a more fine-grained geographic unit than the Census tract, the Census Dissemination Area (DA). While most Riverside residents consider their neighbourhood to be a fairly cohesive community, we explore the effect of socio-economic variation between these small geographic areas on individuals' self-reported health, net of individual level determinants. METHODS: We merge data from the 2001 Census for Riverside, Calgary with a 2004 random telephone survey of Riverside residents. Our data are unique in that we have information on individuals from every DA wholly contained in the Riverside community. These data enable us to conduct multinomial logistic regression analyses of self-reported health using both individual-level and DA-level variables as predictors. RESULTS: We find significant variation in measures of DA socio-economic status within the Riverside community. We find that individual self-reported health is affected by variation in an index of DA-level socio-economic disadvantage, controlling for individual variation in gender, age, and socio-economic status. We investigate each aspect of the DA index of disadvantage separately, and find that average education and the percent of households that are headed by a lone parent are most important. CONCLUSIONS: These findings demonstrate that, even within a cohesive community, contextual effects on health can be located at a smaller geographic level than the Census tract. Research on the effects of local area socio-economic disadvantage on health that combines administrative and survey data enables researchers to develop more comprehensive measures of social and material deprivation. Our findings suggest that both social and material deprivation affect health at the local level.