OBJECTIVE: To use pharmacy and laboratory data to assess diabetes care within a medical group and between medical groups and to determine dispensing patterns and the extent to which providers change therapy based on HbA1c results. RESEARCH DESIGN AND METHODS: Participating groups submitted 1 year of data for continuously enrolled patients. Required data included date of birth, all diabetes-specific prescriptions (oral hypoglycemic agents and insulin), date of prescription, National Drug Code, all HbA1c values, lower and upper normal limits, and date of testing. RESULTS: Few changes in therapy were noted despite the large percentages of patients with suboptimal control. Nearly 90% of the patients treated with medications received a monotherapy regimen involving one of three therapeutic agents: sulfonylureas, metformin, or insulin. More than three-fourths of the patients remained on the same therapy during the observation period despite the fact that 27% of these patients had HbA1c values > or = 8%. Nearly one-fifth (18%) of patients had an HbA1c level of > or = 8% and no further testing for at least 90 days after the "actionable" HbA1C result was obtained. Furthermore, 54% of patients with actionable HbA1c results did not have a change in therapy initiated after the result was available. CONCLUSIONS: The American Diabetes Association recommendations to act on HbA1c values > or = 8% and to follow up regularly on patients found to be in suboptimal control do not appear to be applied in a consistent manner based on the pharmacy and laboratory data analyzed in this sample.