We examined the relationship between the patient activation measure (PAM) and future diabetes-related health outcomes through retrospective analysis of secondary data using multivariate logistic regression. PAM scores from a 2004 survey on 1180 randomly sampled adults with diabetes and health information from a 2006 diabetes registry were the data sources used. The PAM was predictive for hemoglobin A1c (HgA1c) testing (P < .008), low-density lipoprotein cholesterol (LDL-C) testing (P < .005), HgA1c control (P < .01), and all-cause discharges (P < .03), but not for lipid-lowering drug use, LDL-C control, or acute myocardial infarction discharges. These results suggest that PAM scores can be used to identify patients at risk for poorer health outcomes.