OBJECTIVE: To explore the association between change in physical activity levels and fruit and vegetable consumption and changes in self-reported overall health and mental health, of residents living in deprived English communities. DESIGN: Household survey conducted in 2002 and repeated in 2004. SETTING: Thirty-nine deprived UK communities in areas participating in the New Deal for Communities (NDCs), a major government funded community development initiative. PARTICIPANTS: Ten thousand four hundred and nineteen residents in NDC areas and neighbouring comparator areas. MAIN RESULTS: Overall levels of physical activity and fruit and vegetable consumption are low but a large positive change in diet or levels of physical activity is associated with a significant change in mental health (2.86 and 2.71, respectively: P < 0.01). Smaller, but also statistically significant, changes were found in physical health (0.07 and 0.05, P < 0.01). Specific dimensions of mental health which showed a large change in association with lifestyle change were those associated with 'peacefulness' and 'happiness'. CONCLUSIONS: These findings suggest that, for residents of these neighbourhoods, positive lifestyle changes such as increasing physical activity levels and increase in fruit and vegetable consumption are associated with positive changes in mental health. WHAT THIS PAPER ADDS: What is already known? Mental health, a key area of health inequality is related to physical health, and associated with education, employment, environment and community issues. There is known to be a relationship between improved lifestyle (increased physical activity levels and better diet) and better health. What does this study add? This study shows that over two years, measurable changes in lifestyle were associated with improvements in both mental health and self-reported overall health. The association of lifestyle changes with overall health, although statistically significant, were less significant than those with mental health over the same period, suggesting those wanting to measure the impact of community activity on health will be more likely to see a measurable short-term impact on mental, rather than overall self-reported health.