The survival rate for head and neck squamous cell carcinoma remains poor despite therapeutic advances over the last two decades. For patients with disease confined to the head and neck, there are two major and biologically distinct patterns of treatment failures after definitive therapy: recurrence of primary disease and development of second primary tumors. Understanding the biological basis of patterns of treatment failure after definitive therapy is needed to guide the development of adjuvant treatment and strategies to prevent second primary tumors. To determine whether expression of the p53 protein has prognostic significance and/or is associated with patterns of treatment failure, we examined protein expression in primary tumor specimens of patients with head and neck squamous cell carcinoma. Immunohistochemical analysis with a monoclonal antibody (D07) specific for p53 protein was used to detect expression of the protein in formalin-fixed, paraffin-embedded tumor samples from 69 head and neck cancer patients treated with definitive local therapy (surgery and/or radiotherapy) between January 1980 and October 1983 at The University of Texas M. D. Anderson Cancer Center. We quantitated p53 protein expression and assessed its association with duration of patient survival, patterns of treatment failure (recurrence of primary tumor and development of second primary tumor), and other clinical parameters. All reported P values resulted from two-sided statistical tests. We found detectable levels of p53 protein expression in the tumor cell nuclei of 41 of 69 patients. Thirty-six (52%) of 69 patients whose tumors exhibited p53 protein expression in greater than or equal to 10% of the cell nuclei were grouped as p53 positive, and 33 (48%) of 69 patients whose tumors exhibited less than 10% nuclear expression were grouped as p53 negative. The clinical characteristics of the patients in the p53-positive and p53-negative groups were well balanced. Overall survival was significantly lower, and the times to tumor recurrence, to second primary tumors, and to any treatment failure were significantly shorter in the p53-positive group than in the p53-negative group ( P =. 0002, P = .047, P = .003, and P = .0009, respectively), mainly because the p53 positivity was associated with earlier development of tumor recurrence and second primary tumors. The rate of second primary tumor development per person per year was also significantly higher in the p53-positive group than in the p53-negative group. By use of multivariate analysis according to the Cox regression model, p53 expression status was identified as the most significant predictor of overall survival duration ( P = .007), time to tumor recurrence ( P = .053), time to second primary tumors ( P = .035), and time to any treatment failure ( P = .004). The expression of p53 protein in primary head and neck squamous cell carcinoma was significantly predictive of shorter survival because of its association with earlier development of both tumor recurrence and second primary tumors. Thus, p53 expression may be a valuable marker for identifying individuals at high risk of developing a recurrence of primary disease and second primary tumors who may benefit from adjuvant therapy and chemoprevention after definitive local therapy. [J Natl Cancer Inst 1996; 88: 519–29]