The present study was designed to investigate whether there is any correlation between the anticarcinogenic efficacy of selenium (Se) compounds and tissue Se retention under high levels of supplementation. With the use of the dimethylbenz[a]anthracene-induced mammary tumor model in chemoprevention experiments, our data showed that selenomethionine was not as active as selenite over a graded dose range from 1 to 5 p.p.m. Se. Tissue Se concentrations in blood, liver, kidney and skeletal muscle were always higher in rats given selenomethionine compared with those given selenite at each of the three levels tested (1, 3 and 5 p.p.m. Se). The difference was only minimal in blood, but became more pronounced in the liver and kidney, and was quite dramatic in the skeletal muscle. Thus a high tissue concentration or total body burden of Se is not necessarily an indicator of reduced susceptibility to carcinogenesis. The bioavailability of the Se pool in maintaining liver glutathione peroxidase activity during a period of Se deprivation, following excess selenite or selenomethionine loading, was also assessed. The half-life of decay of the enzyme was calculated to be 4.2 and 9.1 days respectively, in those rats that had already been exposed to 3 p.p.m. Se as either selenite or selenomethionine. From a nutritional viewpoint, selenomethionine may be superior to selenite, especially with respect to maintenance of glutathione peroxidase during periods of Se inadequacy, but the reverse seems to be true in terms of anticarcinogenic potency under high levels of Se supplementation. These results suggest that the nutritional and anticarcinogenic efficacies of a given Se compound may not be parallel to each other.