The survival of 1184 British Columbian women whose primary breast cancers were diagnosed and assayed for estrogen receptor (ER) between 1975 and 1981 was studied. Median follow-up was 60 months. ER concentrations yielded greater prognostic information than simple positive and negative categories. When ER data were divided into four strata: less than or equal to 1, 2-9, 10-159 and greater than or equal to 160 fmol/mg cytosol protein, the association of higher ER with prolonged survival was highly significant (P less than 0.0001) and independent of TNM stage, nodal status and menopausal status. ER less than or equal to 1 and ER = 2-9 groups were distinct with respect to overall disease-specific survival. Patient age did not predict survival when controlled for ER. Prolonged recurrence-free survival was associated with higher ER (P = 0.0001) for at least 5 years after diagnosis. This significant trend persisted after adjustments for nodal status, TNM stage, menopausal status and the type of systemic adjuvant therapy.