In an immunohistochemical study of 490 primary breast cancer patients with a follow-up period of more than 10 years, we found that p53 was not a prognostic factor for disease-free or overall survival among the whole cohort or among lymph node-positive or -negative patients. In a multiple logistic regression model classical histopathological parameters, such as lymph node status, number of mitoses, histological grade, and absence of progesterone receptors, were independent, poor prognostic predictors. In univeriate analysis p53 immunoreactivity was positively correlated with the absence of tubule formation, high histological grade (poor differentiation), absence of estrogen receptors (ER), and a high proliferating cell nuclear antigen (PCNA) score (ie, parameters indicative of an aggressive phenotype). The lack of prognostic significance may be attributable partly to the method used, because immunohistochemistry underdetects rather than overdetects p53 protein. No correlation between p53 and c-erbB-2-oncoprotein was demonstrated.