We performed p53 immunostaining in 82 invasive breast carcinomas by using two commercially available antibodies, one of which (DO7) was employed in formalin-fixed paraffin-embedded sections. The other antibody (PAb1801) was evaluated in corresponding acetone-fixed cryostat sections. A greater percent of cases were immunostained with DO7 compared to PAb1801 (52% vs 33%); however, the staining was more often heterogeneous (6-50% cells positive) or focal (< or = 5% cells positive) with DO7 (9% vs 31%). To investigate the genetic relevance of p53 immunostaining, single-strand conformational polymorphism (SSCP) analysis and DNA sequencing were performed on exons 2-11 by using archival tissue samples of 18 cases that were selected on the basis of certain immunostaining patterns. Two (33%) of six tumors with negative staining for DO7 had gene sequence mutations; however, one of these mutations was a base-pair deletion that caused a reading-frame shift and the other was a base-pair insertion that resulted in a stop codon. Both of these tumors exhibited immunostaining with PAb1801, although it was weak and cytoplasmic in one case. Conversely, three (30%) of 10 tumors showing immunoreactivity in 6-100% of cells with both reagents lacked a gene sequence mutation. Of the remaining seven tumors that were positive by SSCP, six contained a point mutation resulting in a base-pair substitution. Despite repeat analyses, one of the cases positive by SSCP failed to demonstrate a mutation in the sequenced exons. Four (80%) of five cases with heterogeneous DO7 immunoreactivity (that is, 6-50% of nuclei positive) were positive for gene sequence mutation. Neither of two cases showing focal DO7 nuclear staining in < 5% of tumor cells contained a mutation in the sequenced exons, and neither of these cases was strongly positive with PAb1801. Staining for either antibody was significantly associated with adverse outcome, as determined by disease recurrence at 52 months median follow-up (DO7, p = 0.01; and PAb1801 p = 0.002, chi-squared test). We conclude that a variety of factors may account for discrepancies when immunohistology is used to evaluate p53 status. These include fixation artifacts, differing epitope specificities of monoclonal reagents, presence of immunohistologically "silent" mutations and, possibly, aberrant overexpression of wild-type protein.