The proliferative activity is one of the most important single prognostic parameters in breast cancer diagnosis and the time-honored measure of proliferative activity, the mitotic figure count, is an integral component of most combined prognostic scores. The detection of the cell cycle-specific antigens Ki-67, and the development of anti-Ki-67 antibodies, including the paraffin-reactive antibody MIB-1, have established immunohistochemical detection of cell cycle-specific antigens as a measure of proliferative activity in breast cancer diagnosis. The current study was performed to correlate mitotic figure counts with the proliferative activity as assessed by MIB-1 immunohistochemistry, taking into consideration the interobserver reliability of 5 pathologists in estimating mitotic figure counts. In 32 consecutive invasive ductal breast carcinomas, mitotic figure counts were performed independently by 5 pathologists. Mitotic activity was expressed as number of mitotic figures per 10 high-power fields and in a 3-tier score according to the Scarff Bloom Richardson system. Immunohistochemistry was performed using MIB-1 antibody, heat-induced epitope retrieval, and the standard avidin-biotin-immunoperoxidase method. MIB-1 immunohistochemistry was assessed in 3 representative 20x fields by semiquantitative estimation (% of tumor cells positive) and by image analysis (number of MIB-1-positive cells/mm2). We found a high degree of interobserver correlation among 4 experienced pathologists, in both mitotic figures per 10 high-power fields and the 3-tier scoring system. We observed significant, albeit weak, correlations between semiquantitative and quantitative MIB-1 immunohistochemistry and the number of mitotic figures per 10 high-power fields (r between .36 and .53), but this significance was lost in 3 of the 5 observers when mitotic activity was expressed in the 3-tier scoring system. This study confirms mitotic figure counting in the hands of experienced pathologists as a valid, reproducible means of assessing proliferative activity in routine breast cancer diagnosis. The statistically significant, albeit only weak, correlation with MIB-1 immunohistochemistry is in agreement with results obtained by others and suggests that MIB-1 immunohistochemistry cannot be translated by a simple conversion factor into combined prognostic scores to replace the time-honored mitotic figure counts.