Analysis of epidermal growth factor receptor (EGFr) and estrogen receptor (ER) was performed on tumor samples from 231 patients with operable breast cancer followed for up to 6 years after surgery. The median duration of follow-up in patients still alive at the time of analysis was 45 months. Thirty-five percent of patients (82) had tumors greater than 10 fmol/mg of 125I-EGF binding (EGFr+) and 47% (109) had cystolic ER concentration greater than 5 fmol/mg (ER+), with a marked inverse relationship between EGFr and ER (P less than .00001). EGFr was second only to axillary-node status as a prognostic marker for all patients in terms of both relapse-free and overall survival in univariate analysis (P less than .001, log-rank EGFr + v EGFr-). For patients with histologically negative axillary nodes, EGFr was superior to ER in predicting relapse and survival (P less than .01 and P less than .005, respectively, compared to P less than .1 and P less than .1, log-rank). In a multivariate (Cox model) analysis, only EGFr--out of EGFr, ER, size, and grade--was predictive for either relapse-free or overall survival for patients with node-negative disease (P = .052 and P = .026, respectively). One hundred eighty-seven case patients in the series were assessed for neu expression immunochemically, and 31 were positive. There was a highly significant increased risk of relapse and death in the positive group. In patients with otherwise good prognostic markers (ER+, node-negative, well-differentiated tumors), neu expression predicted for significantly worsened overall survival.(ABSTRACT TRUNCATED AT 250 WORDS)