To assess the validity of the selenium concentration in human toenails as a measure of selenium intake and to determine other correlates of toenail selenium level, the authors examined the predictors of toenail selenium within two subgroups of a large cohort study of US women. Mean toenail selenium was higher among 38 consumers of selenium supplements (0.904 micrograms/g, standard deviation (SD) 0.217) than among 96 nonusers (0.748 micrograms/g, SD 0.149; p less than 0.001), and a dose-response relation was observed among supplement users (Spearman's r = 0.32; p = 0.05). In a second subgroup of 677 women, selenium supplement use was also associated with higher mean toenail selenium (0.906 micrograms/g, SD 0.214, among 18 users and 0.801 micrograms/g, SD 0.148, among 659 nonusers; p = 0.02), and the dose-response relation was also significant (Spearman's r = 0.50; p = 0.03). The geographic variation in toenail selenium levels was consistent with the geographic distribution of selenium in forage crops. Toenail selenium declined with age and was significantly reduced among cigarette smokers (mean = 0.746, SD 0.124, among 146 current smokers and mean = 0.817, SD 0.159, among 311 never smokers; p less than 0.001) but was not materially affected by alcohol consumption. A dietary selenium score calculated from a food frequency questionnaire failed to predict toenail selenium level, demonstrating the suspected inability of diet questionnaires to measure individual selenium intake because of the highly variable selenium composition of different samples of the same food.(ABSTRACT TRUNCATED AT 250 WORDS)