We compared immunohistochemical staining by two monoclonal antibodies, PC10 and 19A2, to standard methods for cell proliferation, 3H-TdR or BrdU labeling index (S-LI) and S-phase fraction by flow cytometry (S-flow). One hundred thirty-two breast carcinomas were studied retrospectively using formalin fixed, paraffin embedded archival tissues on which S-LI and S-flow had been obtained originally. Percentages of tumor cells positive with PC10 and 19A2 correlated well (r = 0.736, p < 0.0001), although the mean marking index for 19A2 was lower and closer to reference measurements than the mean PC10 index. Correlations between PC10 or 19A2 vs. S-LI, S-flow or DNA ploidy (DNAI) were significant in a group of 64 tumors obtained between 1988 and June 1992, and poor in another group of 68 tumors obtained between 1985 and 1988, suggesting deterioration of stainability with prolonged storage. Discrimination of faint staining from negative nuclei was difficult on PCNA stained sections. Carnoy fixation did not improve results over those fixed with formalin. S-flow and S-LI predicted relapse-free survival, but PCNA indices did not. We conclude that PC10 and 19A2 immunostaining of formalin or Carnoy fixed archival breast cancer tissue correlated with reference measures of proliferation only in cases of short storage periods. Although statistically significant, levels of correlation were too low to use PCNA indices for prognosis.