The objective of this study was to characterize the pattern of p53 mutations in bladder cancer. The sensitivity and specificity to detect these mutations using clinical material was assessed for the following assays: immunohistochemistry, restriction-fragment-length polymorphism, single-strand-conformation polymorphism, and sequencing. Discrepancies of reported results aimed at the identification of genetic alterations in the p53 gene may be due to differences in methodology, as well as to deficient morphological evaluation of the source of tissue utilized. In order to address these critical issues, we have implemented a novel experimental design that permits analysis by molecular genetics and immunopathology techniques in any given tissue specimen, allowing morphological correlation with genotypic and phenotypic characteristics of the tissue analyzed. Forty-two patients affected with bladder tumors in whom paired normal and tumor tissues were available were used for the present study. Nuclear immunoreactivities were observed in 26 out of 42 bladder tumors analyzed. Abnormal shifts in mobility were noted in 14 of the 42 cases in distinct exons, with one tumor revealing 3 mutations. There was a strong association between p53 nuclear over-expression and 17p LOH, as well as p53 nuclear over-expression and detection of mutations by SSCP and sequencing. According to receiver-operating-curve statistical analysis, the accuracy of detecting p53 mutations by IHC was estimated to be 90.3%. It is our conclusion that, when properly used, this is a highly sensitive and specific method with simple application using clinical material.