Context: Medical groups have invested billions of dollars in outpatient Electronic Medical Records (EMR), but few studies demonstrate a positive impact of EMR-based clinical decision support on clinically important patient outcomes. Objectives: To assess the impact of an EMR-based diabetes clinical decision support system on control of glycated hemoglobin (A1c), blood pressure (BP) and LDL Cholesterol (LDL) in adults with diabetes. Design, Setting, and Participants: The study was conducted from October 2006 to May 2007 in Minnesota. Eleven clinics with 41 consenting primary care physicians (PCP) and these physicians’ 2556 diabetes patients were randomized either to receive or not to receive an EMR-based clinical decision support system designed to improve A1c, BP, and LDL in diabetes patients not at recommended clinical goals at the time of an office visit. Main Outcome Measures: Post-intervention levels of A1c, BP, and LDL controlling for pre-intervention levels. Results: Intervention group PCPs used the EMR-based decision support system at over 62.6% of all diabetes patient visits, and intensified drug therapy at 61.9%, 43.6% and 18.8% of visits with uncontrolled A1c, BP, and lipids, respectively. Intervention group diabetes patients had significantly better A1c (P=.014) and SBP (P=.035) but not LDL (P=.63) relative to patients of PCPs randomized to the control arm of the study based on general linear mixed models with a repeated time measurement to control for clustering. Additional analytic models that adjusted for patient age, gender, and co-morbidity showed similar results. Ninety-four percent of PCPs expressed satisfaction with the intervention, and moderate rates of use persisted for over a year after withdrawal of feedback and incentives to encourage use. Conclusions: An EMR-based diabetes clinical decision support significantly improved glucose control and BP control in adults with type 2 diabetes who were not at recommended clinical goals.