The presence of TP53 gene mutations in breast cancer has been associated with worse prognosis. These mutations interfere with the ability of the p53 protein, a transcription factor, to regulate the expression of target genes. Unlike the wild-type protein, which is rapidly degraded in cells, mutated forms have increased half-life and accumulate in tumor cells. Immunohistochemistry (IHC) is widely used in Brazil in the determination of breast cancer patients' prognosis. However, this technique is not able to detect many altered forms of the p53 protein (false-negative results) and readily detects the accumulation of wild-type p53 (false-positive results) that is associated with non-tumoral processes. For these reasons, we have set out to compare the efficiency of IHC with a molecular technique that detects gene variations at the DNA level in the evaluation of Brazilian patients with sporadic breast cancer. We have used denaturing gradient gel electrophoresis (DGGE) to study the TP53 status in 45 tumors, finding 26 allelic variants, most of them located in exon 4. Comparing the two techniques, IHC showed a false-negative rate of 64% and a false-positive rate of 50%. These results confirm the inability of IHC to correctly detect TP53 status, reason because it should not be routinely used to establish prognosis of breast cancer patients in Brazilian Pathology Laboratories. We recommend the utilization of a screening method, such as DGGE, followed by sequencing of altered exonic fragments to correctly detect TP53 gene variants and establish the prognosis of breast cancer patients.