The cytosolic estrogen receptor (ER) content of the primary tumor was determined by isoelectric focusing in 170 consecutive women with unilateral breast cancer diagnosed in 1977 through 1980. No adjuvant treatment over and above surgery and radiotherapy was given. The overall corrected survival was not significantly (p greater than 0.05) higher in ER-rich (less than 0.1 nmol/g DNA) than in ER-poor tumors, but the median period from recurrence to death was longer in the former (16 months) than in the latter (10 months) group. The difference in disease-free survival (DFS) in favour of the ER-rich tumors achieved its maximum-about 15%-after two years (p less than 0.01). At prolonged follow-up, however, the curves converged and there was no significant difference when the whole six year period of observation was taken into account. In patients without axillary metastases the same pattern emerged, with earlier recurrences in ER-poor tumors and a difference in DFS between the two ER groups at two years (p less than 0.01) which was diminished after five years (p less than 0.05). The ER content provided no significant prognostic information in patients with axillary node metastases or locally advanced disease. We conclude from the present and other available data that the ER content in breast cancer would seem to be an indicator of growth rate rather than of metastatic potential and accordingly a predictor of the pattern of recurrence and length of disease-free survival rather than of long-term survival.