The paper reviews arguments that associations between small area socio-economic conditions and individual health are likely to vary according to the type of health condition considered. We comment on the importance of longitudinal research to examine how far area conditions predict later health outcomes, and also how far area variations in health may result from selective migration processes predicted by health status. Using data for 1996 and 2002, from the National Population Health Survey of Canada, linked to small area data on social and on material deprivation in the area of residence for 6950 survey respondents at the two time points, we report on analyses to address these questions. The area measures of material and social deprivation were previously developed by Pampalon and colleagues at the Institut National de Santé Publique de Québec and related to the dissemination area matching the informant's postal code. The health outcomes considered were restriction of activity due to chronic conditions and psychological distress. Our findings suggest that individuals living in materially deprived areas in 2002 were more likely to be affected by health conditions resulting in restriction of activity. Prevalence of psychological distress was higher in areas with greater social deprivation in 2002. Most of these area differences were attenuated when adjustment was made for individual socio-demographic characteristics. Measures recorded in 1996 of individual characteristics and measures of deprivation for area of residence were used to predict change in health outcomes by 2002. Several individual factors (sex, age group, income, household composition) in 1996 were predictive of later health outcomes. After controlling for these individual characteristics the only significant association between health change and area deprivation was with development of restricted activity, which was more common among people who, in 1996, had lived in areas that ranked moderately high on material deprivation. We also report some evidence for selective migration effects, though these mainly seem to operate indirectly via socio-economic characteristics that predict both health outcomes and differential migration patterns. The clearest evidence on this point is for those in with psychological distress in 1996, who were particularly likely to move to an area which was more socially or materially deprived by 2002. This supports the idea that area differences in psychological distress are partly the result of selective migration effects.