The c-erbB-2 oncogene product in serum (serum c-erbB-2) was measured by an enzyme-immunoassay kit. The 12 U/ml cut-off level was estimated as the mean plus two standard deviations for 250 healthy women. With this cut-off level increased serum c-erbB-2 was found in 12.0% of primary breast cancer cases (n = 25), in 4.9% of non-recurrent breast cancer patients (n = 82), and in 31.4% of patients with recurrent breast cancer (n = 35). In patients with primary and recurrent breast cancer, whose sera were assayed concurrently for serum c-erbB-2, CEA and CA15-3, the positive rates of these markers were fairly similar. However, their combined use significantly increased the sensitivity as compared to the use of any one marker alone.