This report addresses diabetes care in the managed care setting and improvement in care brought about by collaboration between 6 Medicare managed care plans (MCPs) and a Peer Review Organization (PRO). The objective was to improve the quality of care of outpatient diabetes patients provided by primary care physicians through the mutual collaboration of 6 Medicare managed care plans and a Medicare Peer Review Organization. The design involved pre-post intervention trial based on 2 random samples, a baseline sample drawn in 1995 and a remeasurement sample drawn in 1996. Medical records of patients in both samples were reviewed by the PRO to determine provision of 14 quality indicator services over a 1-year period. The setting was 6 Arizona Medicare managed care plans comprising approximately 40% of the Arizona Medicare population. Two random samples were drawn from type 2 diabetes patients continuously enrolled in the same managed care plan for at least 1 year. The intervention was comparative feedback of baseline data by the PRO, enabling each plan to compare itself to any other plan on any or all indicators. Each plan developed and implemented its own intervention in response to the 1995 baseline results. The main outcome measures were mean HbA1c, the proportion of HbA1c values below 8%, and positive change in provision of 14 quality indicator services. At postintervention remeasurement, mean HbA1c values fell from 8.9 +/- 2.2 to 7.9% +/- 2.1, and the proportion of patients with HbA1c values below 8.0% rose from 40% to 61.6%. The proportion of the 14 indicator services provided to patients rose from 35% to 55%. The mean number of physician office visits fell 13% and the number of services provided per visit doubled. We conclude that improving the process of care improves glycemic control. Better outpatient diabetes management in competing, capitated managed care plans is an attainable goal when mediated through a neutral third party such as a PRO.