In a retrospective multicenter study to investigate the correlation between estrogen (ER) and/or progesterone receptors (PgR) in primary breast cancer with patient prognosis, 3118 patients with operable breast cancer (International Union Against Cancer Stages I, II, and III) were investigated from ten hospitals in Japan who underwent surgery from October 1972 to December 1982; 3089 were evaluable. The ER-positive and PgR-positive cancers were found in 56% and 34% of patients, respectively. The positivities decreased as the tumor size increased but were independent on lymph node metastasis. There were no significant differences in relapse-free survival (RFS) in relation to receptor status (median follow-up, 89 months [ER], 84 months [PgR]). However, in patients with four or more positive nodes, those with PgR-positive cancer had a longer RFS. The patients with ER-positive cancer survived significantly longer than ER-negative ones, with the greatest difference seen in those with four or more positive nodes. There was a significantly longer postrelapse survival (PRS) for patients with ER-positive cancer because of the different distribution of the major metastasis and better responses to first-line and subsequent treatments. Cox's multivariate analysis showed that overall survival but not PRS was affected by ER (and more weakly by PgR) because of the longer PRS in patients with ER-positive cancer.