Interventions provide a practical context to test core theoretical tenants. Small Steps for Big Changes was a randomized trial designed to modify self-efficacy beliefs to increase free-living physical activity for one year in adults at risk of developing type 2 diabetes. While many studies have examined psychosocial mediators influencing physical activity, few studies have examined serial mediation. The purpose of this study was to examine the serial mediating effects of 6-month physical activity with 12-month self-efficacy (self-regulatory efficacy or task self-efficacy) on participants' 12-month physical activity levels. Hayes' PROCESS macro was used to run a serial mediation analysis to estimate indirect and direct effects. Adults (N=99) who were overweight and had low physical activity levels (mean age=50.9 years, 69.7% female, mean BMI=31.4 kg/cm2) were randomized to receiving either high-intensity interval training (n=47; HIIT) or moderate-intensity continuous training (n=52; MICT). All participants received the same brief behavioural counselling to enhance their self-efficacy. Self-regulatory efficacy and 6-month physical activity levels were found to individually and sequentially mediate the intervention effects (total indirect effect: [29.80, 13.61-49.07]). Self-regulatory efficacy was found to be a causal mechanism in predicting greater physical activity adherence for those in the MICT compared to the HIIT. Task self-efficacy did not mediate intervention effects. Analytic methods used in this study present an innovative method for theory-testing. In line with the social cognitive theory, past physical activity in series with self-regulatory efficacy, was an important mechanism in predicting physical activity one year following a diabetes prevention program.