Epidermal growth factor receptor (EGFr) levels were analyzed in 140 primary breast cancer specimens by immunohistochemical assay (ICA), competitive binding assay (BA), or enzyme immunoassay (EIA). Thirty-nine of 118 specimens (33.1%) were scored as positive by ICA, 30 of 116 (25.9%; cut-off level 10 fmol/mg protein) by BA, and 31 of 80 (38.9%: cut-off level 5 fmol/mg protein) by EIA. Agreement on EGFr status was 72.3% (68/94) between ICA and BA, 77.0% (57/74) between BA and EIA, and 73.8% (59/80) between EIA and ICA. These discrepancies are based on assay differences and the heterogeneous distribution of cancer cells within specimens. Regardless of the assay method used, EGFr status had a significantly negative correlation with estrogen receptor status. Although EGFr-ICA and BA status had no relationship with prognosis, patients with medium and high EGFr-EIA level tumors (over 5 fmol/mg protein) had shorter relapse-free periods than those with low level tumors. However, the prognostic value of positive EGFr-EIA status was weaker than that of c-erbB-2 overexpression.