The value of estrogen receptor (ER) measurements for predicting recurrence and survival rates in primary breast cancer was examined in 121 women who were followed from 5 to 12 years after mastectomy with a median follow-up of 64 months. The prognostic significance of the ER status was evaluated independently and in association with tumor stage, axillary node metastasis, and histopathologic grade. The independent evaluation demonstrated no statistically significant difference in prognosis between women with ER-negative and ER-positive cancers, although the latter group tended to have a longer time to recurrence and longer survival. Multivariate analysis of the data by Cox's proportional hazard regression techniques revealed a synergistic effect of ER status on the risk associated with axillary node metastasis. Patients with nodal metastasis were at 2.8 times the risk of recurrence compared to patients without metastasis. For women with nodal metastasis whose primary cancer was ER-negative, this risk increased to 4.6 times compared to women without metastasis and ER-positive tumors (P = 0.0003). The risk of cancer-related death was 5.6 times more likely for poorly differentiated tumors than for highly differentiated tumors. Patients with poorly differentiated ER-negative tumors were at an even higher risk (7.0) of dying than women with highly differentiated ER-positive carcinomas (P = 0.009). In conjunction with tumor stage, axillary node metastasis and histopathologic grade ER determination is useful for identifying subpopulations at increased risk of tumor recurrence or mortality.