Pretreatment information, type of treatment, and longitudinal follow-up on 1,971 patients with operable breast cancer were used to establish a breast cancer natural history data base (NHDB). Data were available for 957 patients with stage I (node-negative) breast cancer and 1,014 stage II (node-positive) patients. In women with negative nodes, information was available on 759 patients treated at the Milan National Cancer Institute and 188 patients treated at the Royal Marsden Hospital. After adjustment for differences in the distribution of patient prognostic factors, relapse-free survival and overall survival were not significantly different. Of the 1,014 node-positive patients, 540 were treated at the Milan National Cancer Institute, 258 at the Royal Marsden, and 216 at the M. D. Anderson Hospital. Relapse-free survival and overall survival did not significantly differ between Milan patients and those treated at the Royal Marsden Hospital. However, M. D. Anderson Hospital patients did have significantly better relapse-free and overall survival. In each institution, outcome was consistently most dependent on the number of involved axillary lymph nodes and tumor size. Also, similar patterns of survival were observed for each of the institutions. The development of an NHDB can be of value in the identification and evaluation of consistency of prognostic factors, permitting improved comparisons between clinical trials. The development of such a natural history data base (NHDB) provides a reference for assessing the impact of different adjuvant chemotherapy programs, and aids in the design of new protocols.