A simple Macintosh-based image analysis system using consumer-level software was applied to quantitative immunocytochemistry. In three representative x20 fields of 26 infiltrating breast cancers, the number of nuclei were counted that were immunostained with the antibody MIB-1, directed to the cell cycle-specific nuclear antigen Ki-67. The computer-generated proliferative rate exhibited a high linear correlation with visually estimated (r(2) = 0.852, (p < 0.0001) and manually counted proliferative rates (r(2) = 0.643,p < 0.0001). The computer-generated proliferative rate correlated significantly with flow cytometric S-phase fractions (r(2) = 0.345 and r(2) = 0.402, without and with cytokeratin gating, respectively; p < 0.0001) and with histologic and nuclear tumor grades (p < 0.005 and p < 0.039, respectively). No statistically significant interrelationship was found with other prognostic parameters (tumor type, tumor size, hormone receptor expression, overexpression of c-erb B-2 oncogene or of the tumor suppressor gene product p53). This study confirms the interrelationships observed in previous studies between proliferative rate of breast cancers and other prognostic parameters. It presents a Macintosh-based image analysis system as an inexpensive simple tool for the accurate, transparent, and reliable quantification of immunostained nuclei in routine diagnostic immunocytochemistry.