OBJECTIVE: To examine changes in the management of patients with diabetes from 1994 to 1999 using the claims-based Diabetes Quality Improvement Project (DQIP) accountability measures. RESEARCH DESIGN AND METHODS: Administrative claims from an employer-based health insurance cohort in Maine were used to describe the prevalence of claims-based DQIP accountability measures-HbA(1c) testing, dilated eye examination, lipid profile, and monitoring for diabetic nephropathy-from 1994 (n = 1151) to 1999 (n = 2221) in a 100% sample of adults (18-64 years of age) with diabetes. The Mantel-Haenszel chi(2) test for trend was performed on each measure. Prevalence estimates were also stratified by three insurance products: health maintenance organization (HMO), point of service, and indemnity. RESULTS: There was a positive trend for all outcome measures (P < 0.001). The baseline and final frequencies (percent increase) for lipid testing, HbA(1c), dilated eye examination, and screening for diabetic nephropathy were as follows: 13-50% (257%), 37-69% (92%), 30-46% (53%), and 37-50% (36%), respectively. Individuals with diabetes and indemnity insurance were much less likely to receive these measures than individuals with other types of insurance, whereas people in HMOs were more likely to receive HbA(1c) testing and lipid profiles. CONCLUSIONS: The proportion of patients with diabetes receiving DQIP accountability measures significantly increased from 1994 to 1999. There is large variation in prevalence among these measures and insurance products. It is urgent to identify effective mechanisms for delivering consistent preventive care that are congruent with defined standards of benefit.