Mutations in the p53 tumor suppressor gene (also known as TP53) have been detected in a wide variety of human cancers. In breast cancer, the presence of p53 gene alterations has been associated with worse prognosis. We compared a complementary DNA (cDNA)-based sequencing method and an immunohistochemical (IHC) method for their abilities to detect p53 mutations in breast cancer specimens. In addition, we determined the prognostic value of information obtained when these two methods were used. Specimens from 316 primary breast tumors were evaluated for the presence of mutant p53 protein by use of the mouse monoclonal antibody Pab 1801 (that recognizes both wild-type and mutant forms of p53) and standard IHC methods. In addition, the entire coding region of p53 genes expressed in these tumors was screened for mutations by combining reverse transcription, the polymerase chain reaction, and DNA sequencing. Probabilities for overall survival (OS), breast cancer-corrected survival (BCCS; death from breast cancer is the considered event), and relapse-free survival (RFS) were estimated by use of the Kaplan—Meier method, and survival curves for different patient subgroups were compared by use of the logrank method. All reported P values are from two-sided tests. Sixty-nine (22%) of 316 tumors had p53 gene mutations detected by the cDNA-based sequencing method; only 31 (45%) of these mutations were located in evolutionarily conserved portions of the p53 coding region. Sixty-four tumors (20% of the total) had elevated levels of p53 protein as detected by IHC, suggesting the presence of mutations. Of the sequencing-positive tumors (i.e., p53 mutant), 23 exhibited negative IHC reactions, indicating that IHC failed to detect 33% of the mutations. Furthermore, 19 of the IHC-positive tumors were sequencing negative (i.e., p53 wild-type), suggesting a 30% false-positive frequency with IHC. Four tumors (1.3% of the total) could not be analyzed by the cDNA-based sequencing method, and three tumors (1% of the total) could not be analyzed by IHC. The 5-year estimates for RFS, BCCS, and OS were significantly shorter for patients with p53 sequencing-positive tumors than for patients with sequencing-negative tumors ( P = .001, P = .01, and P = .0003, respectively). Patients with IHC-positive tumors showed reduced survival in all three categories when compared with those with IHC-negative tumors, but the differences were not statistically significant. Use of a cDNA-based sequencing method to determine the status of the p53 gene in primary breast cancers yielded better prognostic information than IHC performed with the Pab 1801 monoclonal antibody. [J Natl Cancer Inst 1996;88:173–82]