Introduction: Formal didactic continuing medical education (CME) is relatively ineffective for changing physician behavior. Diabetes mellitus is an increasingly prevalent disease, and interventions to improve adherence to clinical practice guidelines (CPGs) are needed. Methods: A stratified, cluster-randomized, controlled trial design was used to evaluate the effects of a teleconferenced educational detailing (TED) CME on glycemic control (hemoglobin [Hb] A[subscript 1c]) and family physician adherence to national diabetes guidelines. TED employed sequential, small-group, case-based education using CPGs delivered by a diabetes specialist. Medical record audit data from baseline through the end of a 12-month postintervention period were compared for the control and intervention groups. Satisfaction with the intervention was evaluated. Results: Sixty-one physicians provided 660 medical records. The intervention did not affect mean Hb A[subscript 1c] levels but did significantly (p = 0.04) alter the distribution of patients by category of glycemic control, with fewer in the intervention group in inadequate control (15.8% versus 23.9%). More patients took insulin (alone or with oral agents) in the intervention group (21.2% versus 12.0%, p = 0.03), and more took oral agents only in the control group (89.0% versus 82.9%, p = 0.005). More patients in the intervention group had documentation of body mass index (7.8% versus 1.9%, p less than 0.02), eye exam (12.1% versus 5.1%, p = 0.02), and treatment plan (43.5% versus 23.6%, p = 0.01) and used a flow sheet (14.6% versus 7.7%, p less than 0.03). Although there was general satisfaction with the teleconferencing format, specialist educators found the format more challenging than the family physicians. Discussion: CME delivered by teleconference was feasible, well attended, well received by participants, and improved some key diabetes management practices and outcomes.