Both goldthioglucose (GTG)-treated and the genetically obese (C57BL/6J ob/ob) mice were fed semisynthetic diets containing either soy protein isolate (SPI) or casein as a protein source, or laboratory chow. In GTG-induced obese mice, the plasma cholesterol level corre lated positively with their body weight. The level was highest in mice fed high-fat diet, and lowest in ones fed laboratory chow. No difference was observed between SPI and casein groups whether the diet was low-fat or high-fat. Thus, in the GTG-treated mice, SPI did not have a hypocho lesterolemic effect while dietary fat had a hypercholesterolemic effect, and laboratory chow contained some component(s) which can lower the plasma cholesterol level. Both neutral and acidic steroid contents in feces of the SPI group were not different from those of the casein group, and both groups of mice excreted a smaller amount of steroids than mice fed laboratory chow. Results of essentially the same tendency were obtained with normal mice regarding the effects of SPI and casein, although the degree of hypercholesterolemia was lower in high-fat-fed normal mice than in similarly fed GTG-treated mice. These results strengthened the inverse correlation between the amount of fecal steroids and the plasma cholesterol level upon feeding various proteins, indicating that the former is one of the important factors that determine the latter. The ob/ob mice showed a marked hypercholesterolemia irrespective of the kind of diet. The amount of fecal steroids was highest in the laboratory chow group and lowest in the casein group. This indicates that some factor(s) other than fecal steroid excretion is dominantly responsible for their hyper cholesterolemia.