AIMS: This study aimed to investigate whether an established behavioural intervention, Walking Away from Type 2 Diabetes, is effective at promoting and sustaining increased walking activity when delivered within primary care. METHODS: Cluster randomized controlled trial involving 10 general practices recruited from Leicestershire, UK, in 2009-2010. Eight hundred and eight (36% female) individuals with a high risk of Type 2 diabetes mellitus, identified through a validated risk score, were included. Participants in five practices were randomized to Walking Away from Type 2 Diabetes, a pragmatic 3-h group-based structured education programme incorporating pedometer use with annual follow-on refresher sessions. The primary outcome was accelerometer assessed ambulatory activity (steps/day) at 12 months. Longer term maintenance was assessed at 24 and 36 months. Results were analysed using generalized estimating equation models, accounting for clustering. RESULTS: Complete accelerometer data for the primary outcome were available for 571 (71%) participants. Increases in ambulatory activity of 411 steps/day [95% confidence interval (CI): 117, 704] and self-reported vigorous-intensity physical activity of 218 metabolic equivalent min/week (95% CI: 6, 425) at 12 months were observed in the intervention group compared with control; differences between groups were not sustained at 36 months. No differences between groups were observed for markers of cardiometabolic health. Replacing missing data with multiple imputation did not affect the results. CONCLUSIONS: A pragmatic low-resource group-based structured education programme with pedometer use resulted in modest increases in ambulatory activity compared with control conditions after 12 months when implemented within a primary care setting to those at high risk of Type 2 diabetes mellitus; however, the results were not maintained over 36 months.