Most studies of selenium and thyroid hormone have used sodium selenite in rats. However, rats regulate thyroid hormone differently, and selenite, which has unique pharmacologic activities, does not occur in foods. We hypothesized that selenium in food would have different effects in humans. Healthy men were fed foods naturally high or low in selenium for 120 d while confined to a metabolic research unit. Selenium intake for all subjects was 47 microg/d (595 nmol/d) for the first 21 d, and then changed to either 14 (n = 6) or 297 (n = 5) microg/d (177 nmol/d or 3.8 micromol/d) for the remaining 99 d, causing significant changes in blood selenium and glutathione peroxidase. Serum 3,3',5-triiodothyronine (T3) decreased in the high selenium group, increased in the low selenium group, and was significantly different between groups from d 45 onward. A compensatory increase of thyrotropin occurred in the high selenium group as T3 decreased. The changes in T3 were opposite in direction to those reported in rats, but were consistent with other metabolic changes. By d 64, the high selenium group started to gain weight, whereas the low selenium group began to lose weight, and the weight changes were significantly different between groups from d 92 onward. Decreases of serum T3 and compensatory increases in thyrotropin suggest that a subclinical hypothyroid response was induced in the high selenium group, leading to body weight increases. Increases of serum T3 and serum triacylglycerol accompanied by losses of body fat suggest that a subclinical hyperthyroid response was induced in the low selenium group, leading to body weight decreases.