The objective of this study was to analyze the impact of a community-based physical activity program, the Health Academy Program (Academia da Saúde or PAS), on leisure-time physical activity (LTPA) among the population living at the Brazilian state capitals. We pooled individual data from the National Surveillance for Protective and Risk Factors for Chronic Diseases (VIGITEL) between 2006 and 2016 and estimated odds ratios according to levels of exposure by using a multilevel logistic regression. Total sample was composed of 572,437 individuals. In the initial model, chances of reaching sufficient LTPA was 1.20 (95%CI: 1.16-1.25) times higher among individuals exposed since 2011. In the analyses adjusted for year, sex, age and education, this probability was only 1.04 (95%CI: 1.00-1.08) times higher among exposed individuals. Odds of reaching recommended LTPA was 1.09 (95%CI: 1.04-1.15) times higher among women exposed since 2011 as compared to women in the control group with no exposure. No other statistically significant results were found. We conclude that the PAS cannot substantially affect whole populations. Yet it is possible to visualize a positive influence of the program on specific subgroups, pointing to its potential to reduce gender inequity in LTPA practice. We recommend more tailored interventions before indistinctively scaling up the program, as well as we suggest better monitoring for large scale evaluations.