A case-control study of oral cancer was conducted in western Washington state between 1983 and 1987. Cases (n = 379) were identified through a population-based registry, and controls (n = 514) were selected by telephone using random digit dialing. Subjects participated in a personal interview, completed a food-frequency questionnaire, and submitted clippings from the nails of each great toe for the determination of selenium and zinc concentrations. The odds ratio (OR) for low selenium levels in nail tissue (lowest 25% of the distribution compared to the upper 75%) was 1.4 (95% confidence interval (CI) 1.0-2.2). Likewise, the odds ratio for low zinc levels in nails was 1.6 (95% CI 1.0-2.3), but for low dietary zinc was 1.0 (95% CI 0.7-1.7). Men with oral cancer had lower nail selenium levels than did the controls (OR = 1.9), but women with oral cancer did not (OR = 0.6). Individuals 20 to 39 years of age with oral cancer, in particular, were more likely to have lower selenium levels in nail tissue than controls (OR = 16.4). There was a significant interaction between selenium and ascorbic acid levels which could not be explained by cigarette use. Subjects at greatest risk had low levels of both nutrients (OR = 3.8 for smokers and OR = 5.7 for non-smokers). However, since the elements were deposited in the nail matrix close to the date of diagnosis, the differences in the element concentrations between cases and controls may have been a result of the disease. Further etiologic studies of selenium, vitamin intake and oral carcinoma are warranted.