This paper uses national survey data for young adults in England to explore empirically the relationships between social fragmentation in communities (measured for geographical areas), social support experienced by individuals from their immediate social circle, and psychosocial health of young adults. After reviewing previous research about these associations, we adopted an empirical approach to these questions, which was innovative in using data on area social fragmentation from a different source to the survey data on individuals. Also, we have examined the relevance for mental health of interactions between individual social support and area social fragmentation, as well as their independent associations with health. To test these ideas empirically, we present a statistical analysis, using survey data from the national Health Survey for England on young people aged 16-24 years, linked to a geographical indicator of social fragmentation, derived from the population census and with a measure of material poverty. The outcome variable was distress measured by the General Health Questionnaire (GHQ). In a logistic regression model that controls for grouping of individuals within areas we included data on individuals' sex, ethnic group, employment status, social class and educational level. Controlling for these indicators, we demonstrate that risk of individual distress (indicated by GHQ score of 3+) was significantly and positively associated with area social fragmentation and there was a significant association with social support received within the individual's immediate social circle, which was negative ('protective'). An index of material poverty in one's area of residence did not predict individual distress. There was no evidence that social support was more 'protective' in areas of greatest social fragmentation. We also note that while being in employment was associated with better mental health in this sample, higher educational level was associated with worse average levels of distress (controlling for age). We consider some of the policy implications of the findings.