Breast tissue samples, including normal breast, nonmalignant disorders, and breast carcinomas (n = 257), were tested with monoclonal antibody Ki67 to define the growth fraction in each tissue subgroup. Immunocytochemical assays using anti-Ki67 and avidin-biotin-peroxidase complex and/or alkaline phosphatase anti-alkaline phosphatase were applied in frozen sections. The immunoreactions were analyzed with a computerized system of image analysis referred to as SAMBA (Systeme d'Analyse Microphotometrique à Balayage Automatique). This system permitted a multiparametric and automatized analysis of colored images. The results obtained were: (a) the SAMBA analysis of Ki67-positive staining was accurate, reliable, and reproducible; (b) the anti-Ki67 immunostaining was significantly (P less than 0.01) increased in malignancies and was related to the tumors' degree of differentiation, the vascular invasion, and the presence of axillary lymph node metastases; (c) anti-Ki67 immunostaining is increased (P less than 0.01) in tumors in which estrogen receptor and progesterone receptor antigenic sites are not detected. It is concluded that the SAMBA analysis of the anti-Ki67 immunocytochemical assay provides relevant information in selecting subgroups of patients with higher risk for relapse.