BACKGROUND. The incidence of head and neck cancer is increasing. To improve the survival of head and neck cancer patients, an effective program of screening and/or chemoprevention of second malignancies is essential. An analysis of the incidence, time to development, and risk factors of second malignant tumors in head and neck cancer patients can contribute to the design of effective screening and chemoprevention programs. METHODS. Eight hundred, fifty-one patients with initial squamous cell carcinoma of the larynx (n = 224), tonsils (n = 189), pyriform sinus (n = 165), oral cavity (n = 129), mobile tongue (n = 72), and base of tongue (n = 72) treated from 1978 to 1990 were analyzed for the presence of a second malignancy after initial therapy. Of these 851 patients, 544 (64%) were documented smokers and 35 (4%) were nonsmokers. No smoking information was available for 272 patients. Four hundred, fifty-four patients (53%) were consumers of alcohol and 64 patients (8%) were nondrinkers. Alcohol consumption information was not available for 333 patients. RESULTS. One hundred, sixty-two (19%) second head and neck carcinomas occurred in the original 851 patients. Sixty-six patients (41%) had synchronous tumors, and 96 patients (59%) had metachronous tumors. The probability of developing a second metachronous cancer 5-years after undergoing treatment for the initial head and neck cancer was 22%. Borderline statistical significance was observed in the 5-year second cancer incidence based on the site of the initial primary cancer (46% for the base of tongue, 34% for the pyriform sinus, 23% for the larynx, 18% for the oral cavity, 15% for the tonsils, and 10% for the mobile tongue). Tobacco smoking (3% for nonsmokers vs. 26% for < or = 20 pack-years vs. 42% for > 20 and < or = 40 packs/year vs. 30% for > 40 packs/year of smoking) and the consumption of alcohol (5% for non-drinkers vs. 32% for drinkers) were both statistically significant in predicting the likelihood of developing a second malignancy. Multivariate analysis revealed that the two independent variables that influenced the occurrence of a second metachronous cancer were the anatomic site of the original primary cancer and patient age. The survival rate after the second cancer was influenced significantly by the site of the second cancer (20% for a second head or neck cancer, 3% for a second esophageal cancer, and 2% for a second lung cancer). Continued smoking (20% for non-smokers vs. 5% for smokers) and continued alcohol consumption (27% for nondrinkers vs. 6% for drinkers) also adversely influenced the survival after the occurrence of a second cancer. CONCLUSIONS. This study confirms the high rate of second cancers in patients with initial head and neck malignancies. The development of a second malignancy is almost always fatal. Screening programs and chemoprevention trials should be directed toward cancer patients with initial head and neck cancers. Only the small subset of nonsmokers and nondrinkers should be excluded from such trials.