A method of carrying out region-wide audit for breast cancer was developed by collaboration between the cancer registry, providers and purchasers as part of work to fulfill the 'Calman-Hine' recommendations. In order to test the audit method, a retrospective audit in North Thames East compared practice in 1992 against current guidelines. The analysis compared care in specialist and non-specialist centres. A stratified random sample comprising 28% of all breast cancer patients diagnosed in 1992 was selected from the population-based Thames Cancer Registry. The data for 309 patients with stage I-III tumours were analysed by hospital type using local guidelines. No difference between specialist (high volume) and non-specialist centres was detected for factors important in survival. Pathological staging was good with over 70% reporting tumour size and grade. A small number of patients were undertreated; after conservative surgery, 10% (19) of women did not receive radiotherapy, and 15% (8) of node-positive premenopausal women did not receive chemotherapy or ovarian ablation. In contrast, a significant trend with hospital volume was found for several quality of life factors. These included access to a specialist breast surgeon and specialist breast nurses, availability of fine-needle aspiration (FNA), which ranged from 84% in high-volume to 42% in low-volume centres, and quality of surgery (axillary clearance rates ranged from 51% to 8% and sampling of less than three nodes from 3% to 25% for high- and very low-volume centres respectively). Confidential feedback of results to surgeons was welcomed and initiated change. The summary information gave purchasers information relevant to the evaluation of cancer services. While the audit applied present standards to past practice, it provided the impetus for prospective audit of current practice (now being implemented in North Thames).