Beta-carotene has established efficacy in animal models of oral carcinogenesis and has been shown to regress oral precancerous lesions in humans. The purpose of this study was to see whether these effects extended to the prevention of oral/pharyngeal/laryngeal (head and neck) cancer in humans. The subject population for this randomized, placebo-controlled, double-blinded clinical trial included 264 patients who had been curatively treated for a recent early-stage squamous cell carcinoma of the oral cavity, pharynx, or larynx. Patients were assigned randomly to receive 50 mg of beta-carotene per day or placebo and were followed for up to 90 months for the development of second primary tumors and local recurrences. After a median follow-up of 51 months, there was no difference between the two groups in the time to failure [second primary tumors plus local recurrences: relative risk (RR), 0.90; 95% confidence interval (CI), 0.56-1.45]. In site-specific analyses, supplemental beta-carotene had no significant effect on second head and neck cancer (RR, 0.69; 95% CI, 0.39-1.25) or lung cancer (RR, 1.44; 95% CI, 0.62-3.39). Total mortality was not significantly affected by this intervention (RR, 0.86; 95% CI, 0.52-1.42). Whereas none of the effects were statistically significant, the point estimates suggested a possible decrease in second head and neck cancer risk but a possible increase in lung cancer risk. These effects are consistent with the effects observed in trials using intermediate end point biological markers in humans, in which beta-carotene has established efficacy in oral precancerous lesions but has no effect or slightly worsens sputum cytology, and in animal carcinogenicity studies, in which beta-carotene has established efficacy in buccal pouch carcinogenesis in hamsters but not in animal models of respiratory tract/lung carcinogenesis, with some suggestions of tumor-promoting effects in respiratory tract/lung. If our results are replicated by other ongoing/completed trials, this suggests a critical need for mechanistic studies addressing differential responses in one epithelial site (head and neck) versus another (lung).