Breast-conserving therapy (BCT) has survival results comparable to those obtained with modified radical mastectomy (MST). However, studies suggest variations in the rates of breast-conserving therapy that are not explained by comorbidities or age. The Virginia Health Quality Center collaborated with 40 Virginia hospitals to address patterns of breast cancer treatment. Medicare Peer Review Organization data files were used to identify all Medicare beneficiaries with a primary diagnosis of breast cancer from January 1, 1992, through June 30, 1993. Explicit chart review was performed on a random sample of patients with early stage disease receiving either BCT or MST. At baseline, the BCT rate was 21.1% after controlling for access to radiation facilities, medical conditions, and demographic variables. Hospitals with the highest rates of appropriate BCT served as resources for the project. We provided data on BCT rates to 40 participating hospitals. Hospitals then submitted comprehensive plans to address the performance of BCT. A four-step cooperative improvement intervention was employed to initiate and sustain changes at the hospital level. Interventions included individual hospital feedback, dissemination of model cancer care processes at high-performing institutions, and integration of oncology services for breast cancer treatment decisions. One year after implementation, the overall BCT rate in Virginia increased to 25.5%. Hospitals with the lowest BCT rates increased their average rate from 6.6% to 21.2%. Middle tercile hospitals increased BCT rates by 10%. Variation between the lowest and highest terciles was reduced to 9%. Variation in BCT rates can be reduced by a collaborative program centered on addressing processes of care for breast cancer treatment for Medicare patients with early stage disease.