Using data from a case-control study conducted between 1984 and 1992 in the provinces of Milan and Pordenone, northern Italy, on 439 cases of oral and pharyngeal cancers and 2106 hospital controls, we computed the population attributable risk for oropharyngeal cancer in relation to tobacco, alcohol, and a measure of low beta-carotene intake. Two different models were used for estimating relative risks, one assuming that the three factors act multiplicatively on the relative risk and the second estimating separately each combination of alcohol and tobacco and assuming a multiplicative model only for beta-carotene. The estimated attributable risks were similar for the two models considered. For both models and both sexes, the single factor with the highest attributable risk was smoking, which accounted for 81-87% of oral cancers in males and for 42-47% in females. Alcohol explained about 60% of male cases, but only 15% of female ones, and low beta-carotene accounted for 24% of total cases (25% of males, 17% of females). Together the three factors were responsible for 91-94% of oropharyngeal cancers in males, 51-57% in females, and 85-88% in both sexes combined. The present knowledge of major identified risk factors could, in principle, reduce the burden of the disease in Italy from 2400 to about 200 deaths per year for males and from 500 to 230 for females, thus explaining the difference in incidence and mortality between the two sexes.