To investigate whether or not hormone receptor determination gives independent prognostic information for long term survival of patients with carcinoma of the breast, we studied 1,392 patients with early carcinoma of the breast. Patients were part of two prospective, multi-institutional trials, the first begun in 1974 and the second in 1980. Estrogen receptor assays were performed on all primary specimens taken of the carcinoma of the breast. Initial treatment for all patients was a modified radical mastectomy. Nine hundred and seventeen patients had negative axillary nodes and were observed without additional therapy. Four hundred and seventy-five had positive nodes and were randomized to receive combination chemoendocrine adjuvant therapy. One thousand and sixty-three (76.4 per cent) of the patients were found to have estrogen receptor positive (ER+) tumors (greater than or equal to 3 femtomoles per milligram cytosol of protein). The ten year over-all survival rate of 65.9 per cent was significantly better than that of 329 (23.6 per cent) patients with estrogen receptor negative (ER-) tumors (less than 3 femtomoles per milligram cytosol protein), who had a ten year over-all survival rate of 56.0 per cent (p = 0.0001). Higher estrogen receptor values were associated with Caucasian (p = 0.0001) and postmenopausal patients (p = 0.0001). In a proportional hazards regression model, patients with ER+ tumors had a significantly longer over-all survival period (p = 0.0001), but only a marginally improved disease-free survival time (p = 0.07) when compared with patients who had ER- tumors. These results indicate that ER determination does have prognostic value for long term over-all survival of patients with carcinoma of the breast. The greater importance of ER analysis to over-all compared with disease-free survival may be related to more easily managed recurrent disease among the ER+ group.