Estrogen and progestin receptor levels (ER and PgR) in tumors from 506 patients with primary breast cancer diagnosed in 1979, 1980, and 1981 were measured by a Scatchard plot analysis. At a median follow-up time of 3.5 years the prognostic value of the receptor levels was evaluated and compared with other tumor and patient characteristics. No relation was found between receptor levels and tumor, lymph node, metastasis (TNM) classification or location of the primary tumor. A significant positive rank correlation was observed between ER and PgR levels (rs = 0.57) and between ER level and age of the patients (rs = 0.39, P less than 0.001). The observed association between ER level and menopause status could not be maintained after correction for age. Independent prognostic factors for overall survival were tumor size (P = 0.002), the number of positive lymph nodes (P less than 0.001), age at primary surgery (P less than 0.001), the PgR level of the tumor (P less than 0.001), but not ER level. Independent prognostic factors for relapse were tumor size (P = 0.003), number of positive lymph nodes (P less than 0.001), age (P = 0.006), menopause status (P = 0.02), PgR level (P = 0.007), but not ER level. Finally, for death rate after relapse the following prognosticators were identified: size of the primary tumor (P = 0.03), number of positive lymph nodes (P = 0.03), age (P = 0.003), site of relapse (P less than 0.001), ER level (P = 0.02), and PgR level (P = 0.04). Patients with tumors containing low positive PgR levels (10 to 20 fmol/mg protein) had a slightly better prognosis than patients with PgR-negative tumors. It is concluded that the PgR level of the primary tumor is a better prognosticator than the ER level. The ER offered no additional ability for discriminating between low- and high-risk patients once PgR was included in the model. In contrast, PgR was capable of improving on the discriminating ability of ER. In addition, patients with tumors containing both PgR and ER showed the best prognosis. Therefore, it is recommended that ER and PgR should be assayed in all breast cancer biopsies.