The recent availability of the monoclonal antibody MIB-1 (which is able to detect the human nuclear cell proliferation-associated antigen Ki-67 even on formalin-fixed, paraffin-embedded sections, microwave-treated and routinely processed for immunohistochemistry) could open new avenues for validation of the clinical role of tumour cell proliferation on large, consecutive and unselected series of human tumours. However, the routine use of such a marker requires a methodological standardization as well as the comparative assessment of some technical and biological aspects. The MIB-1 index was determined in parallel samples from 50 consecutive invasive breast carcinomas processed with different fixatives for different times. The median values of MIB-1 indices following 2, 6 and 24 h of formalin fixation were similar (29.4%, 30.6% and 29.7%, respectively) and consistent with those reported in the literature; squared linear regression coefficients were 0.99. The median values of MIB-1 indices were markedly lower in Bouin-fixed, paraffin embedded, and in frozen samples (20.0% and 19.8%, respectively), with a poor correlation coefficient with the values detected following formalin fixation (R2 = 0.456). Moderate and poor correlations were observed between Ki-67 index and MIB-1 detected on frozen (rs, 0.78) or formalin-fixed, paraffin-embedded samples (rs, 0.47) and a minimal concordance was observed between TLI and MIB-1 or Ki-67 (rs, 0.25 and 0.22, respectively). Our results indicate interference of the fixative type on immunoreactivity to MIB-1 and also suggest that Ki-67 and MIB-1 reacted with different epitopes of the same antigen.