This study was designed to test the validity of a transtheoretical model's physical activity (PA) stage measure with intention and different intensities of behavior in a large population-based sample of adults living with diabetes (Type 1 diabetes, n = 697; Type 2 diabetes, n = 1,614) and examine different age groups. The overall "specificity" (classified correctly as inactive) for Type 1 diabetes was 69.3% based on the percentage of individuals in precontemplation, contemplation, and preparation not meeting Canada's PA guidelines. "Sensitivity" (classified correctly as active) was 82.5% based on the proportion of active participants in action and maintenance. In the Type 2 diabetes group, the overall specificity and sensitivity was 63.9% and 88.2%, respectively. No significant differences were found between the diabetes groups for stage distribution patterns, and sensitivity and specificity values. The majority of the study hypotheses related to intention and behaviors were confirmed, providing further supportive evidence for the utility of this PA staging measure for the diabetes population. (Contains 2 figures, 4 tables and 2 footnotes.)