The estimation of the Ki-67 index in human breast cancer tissue has been proven to be a useful prognostic tool. The examination can be performed, however, only on frozen sections (FS). The development of an antibody directed against parts of the Ki-67 antigen (MIB-1) has opened a new route to determine the proliferative activity on paraffin sections (PS). MIB-1 immunohistochemistry is used instead of Ki-67 immunohistochemistry if a tumour is delivered to the pathologist after formalin fixation or if that part of the tissue suspicious for breast cancer must be totally embedded in order to confirm the diagnosis. The present study compares the findings of Ki-67 (FS) and MIB-1 (FS and PS) immunohistochemistry in a total of 544 cases of human breast cancer. The findings confirm a good statistical correlation between the Ki-67 and the MIB-1 findings. The MIB-1 results are 2-2.5 times higher in FS than in PS. Good agreement exists between the Ki-67 indices determined on FS and the MIB-1 indices determined on PS. If the cut-off value for the separation of Ki-67 negative and positive cases is defined as 10%-20%, a MIB-1 index in PS of 10% permits the correct prediction of a negative Ki-67 index in 97% of the cases, and a MIB-1 index of 30% or more correctly predicts a positive Ki-67 index in 90% or more of the cases. Hence, the determination of the MIB-1 index on PS may replace the determination of the Ki-67 index on FS with a high degree of probability.