To evaluate the ability of serum c-erbB-2 protein to (1) indicate occult and manifest metastases and (2) reflect response to first-line therapy, serial serum c-erbB-2 measurements were performed in a retrospective series of 52 primary breast cancer patients who had developed metastatic disease during follow-up. The results were compared with CA 15-3. Preoperatively, 31% (16/52) of the primary breast cancer patients had elevated c-erbB-2 concentrations. The CA 15-3 positivity rate was 13% (7/52). After surgery, 10 of the 52 patients showed either stable but highly elevated or rising c-erbB-2 serum levels indicating serum c-erbB-2 producing minimal residual disease. Increasing CA 15-3 concentrations were seen in only three patients. Elevated serum c-erbB-2 levels predicted manifest metastases in 27 and 50% of the patients at 6 and 3 months, respectively, prior to clinical diagnosis. CA 15-3 was less sensitive. Only 16 and 32% of the patients had increased CA 15-3 serum concentrations at 6 and 3 months, respectively, prior to clinical detection. The positivity rates of c-erbB-2 and CA 15-3 were similar when metastases were clinically diagnosed. Elevated c-erbB-2 concentrations were found in 62% (32/52). The sensitivity of CA 15-3 was 56% (29/52). The association between serum profiles and response to first-line therapy was evaluated in detail for 45 patients. Serial c-erbB-2 and CA 15-3 measurements reflected disease course in 24 and 27 patients, respectively. The serum profiles of c-erbB-2 and CA 15-3 were similar in 17 patients. In summary, our results suggest that serial determinations of serum c-erbB-2 are useful to monitor breast cancer patients.