Mediation analyses in faith-based physical activity (PA) interventions targeting African-American adults are lacking. The purpose of this study was to examine the psychosocial mediators of a faith-based PA intervention with African-American adults. Churches were randomly assigned to receive immediate or delayed (1-year later) training in PA program implementation. A subsample of participants from randomly selected churches took part in telephone surveys at baseline and at 1 year. The primary outcome was percentage of participants meeting PA recommendations. MacKinnon's product of coefficients was used to test for mediation. Participants (n = 418) from 20 churches completed the baseline and 1-year follow-up surveys. There were no statistically significant changes in PA behavior at 1 year. The intervention had a marginally significant effect on increasing the amount of instrumental church support received by church members. However, none of the psychosocial variables tested were found to be significant mediators of the intervention. Mediation analyses provided insight into potential reasons as to why the Health-e-AME intervention did not change PA. The intervention did not successfully change the targeted mediators hypothesized to change PA. Potential reasons for these shortcomings as well as issues to address in future faith-based studies are discussed. (Contains 2 tables and 1 figure.)