Increasing population physical activity (PA) is a public health priority. An approach to increasing PA in an urban community, based on the social ecology model, is presented and evaluated. A quasi-experimental (non-equivalent control group) design was chosen to test whether this approach can increase, significantly, the population proportion that is physically active. Two deprived inner-city electoral ward areas of Sheffield, UK, with similar socio-demographic and health profiles were selected. Implementation was carried out in five phases over 21 months: preparation and piloting, initial survey estimates, community awareness campaign, PA intervention and evaluation. Impact was evaluated by recording uptake and attendance at all sessions, and using a post-intervention postal survey. At follow-up, questionnaires were sent to 2500 randomly selected addresses in both areas, which assessed changes in PA, health and PA participation over the last year. A pilot baseline survey confirmed similar proportions taking some form of PA on most days: intervention 36% (25-47), control 33% (21-45). At follow-up, 38 different activity groups were in place and 1275 individuals had attended at least one activity. Satisfactorily completed responses were received from 1532 (31%), 55% (846) intervention, 45% (686) control. Relative to the control, the intervention sample demonstrated trends towards: being more physically active compared with one year ago, greater readiness to take up PA, better general health and improved health compared with one year ago (p < or = 0.001). Further, 30.6% (intervention) vs 18.3% (control) reported an increase in PA compared with one year ago, while 13.7% (intervention) vs 24.5% (control) reported no intention to exercise. These differences in proportions translated to an overall effect size estimate of 0.23 for the intervention. Relative to those whose PA had remained the same, intervention area residents were more likely to report being more active (odds ratio (95% confidence interval) = 1.79 (1.38-2.32), p < 60; 0.001). These results suggest that, even within one of the most deprived inner-city areas in the UK, a community-based social ecological approach can positively change PA behaviour and attitudes.