In view of current emphasis on identifying prognostic factors for patients with early breast cancer, we studied the importance of tumor size to survival among 1392 patients with primary operable breast cancer who were followed up prospectively. All patients had modified radical mastectomies. Nine hundred seventeen patients had negative nodes and did not receive postoperative adjuvant therapy. Four hundred seventy-five patients had node involvement and received combination chemoendocrine therapy. In a Cox's proportional hazards model, tumor size was a significant predictor of disease-free and overall survival when the number of positive nodes, estrogen receptor status, menopausal status, and race were considered. Among the node-negative group, tumor size explained considerable variation in disease-free and overall survival, varying from a 10-year disease-free and overall survival of 80% and 99% for patients with estrogen receptor-positive tumors measuring 1 cm or less to a 10-year disease-free and overall survival of 51% and 59% for patients with tumors larger than 5 cm.