Background: We examined the effect of power training on habitual, intervention and total physical activity (PA) levels in older adults with type 2 diabetes and their relationship to metabolic control. Materials and Methods: 103 adults with type 2 diabetes were randomized to receive supervised power training or sham exercise three times/week for 12 months. Habitual, intervention, and total PA, as well as insulin resistance (HOMA2-IR) and glycosylated hemoglobin (HbA1c), were measured. Results: Participants were aged 67.9 ± 5.5 yrs, with well-controlled diabetes (HbA1c = 7.1%) and higher than average habitual PA levels compared to healthy peers. Habitual PA did not change significantly over 12 months (p = 0.74), and there was no effect of group assignment on change over time in habitual PA over 0–6 (p = 0.16) or 0–6–12 months (p = 0.51). By contrast, intervention PA, leg press tonnage and total PA increased over both 6- and 12-month timepoints (p = 0.0001), and these changes were significantly greater in the power training compared to the sham exercise group across timepoints (p = 0.0001). However, there were no associations between changes in any PA measures over time and changes in metabolic profile. Conclusion: Structured high-intensity power training may be an effective strategy to enhance overall PA in this high-risk cohort.