Male weanling guinea pigs were fed selenium-deficient and control diets for 10 weeks and the activity of glutathione peroxidase was studied in liver 105,000xg supernatant and in blood using both 0.25 mM H2O2 and 1.5 mM cumene hydroperoxide as substrates. In control liver supernatant, the activity measured with H2O2 was about 20% of that measured with cumene hydroperoxide, indicating that only a small fraction of the total glutathione peroxidase present was selenium-dependent glutathione peroxidase (Se GSH-Px). Se GSH-Px decreased to 20% of control Se GSH-Px in livers of animals fed the selenium-deficient diet, but because it was such a small fraction of the total glutathione peroxidase activity, selenium deficiency did not affect total glutathione peroxidase activity in liver supernatant. In contrast to the liver results, blood glutathione peroxidase activities were higher when H2O2 was used as substrate than when cumene hydroperoxide was used. Gel filtration experiments suggested that this was due to interference with the assay by guinea pig hemoglobin which apparently caused oxidation of NADPH in the assay system. This oxidation was greater when H2O2 was used as substrate than when cumene hydroperoxide was employed. The presence of Se GSH-Px in red blood cells and its decrease in selenium deficiency were demonstrated by the gel filtration experiments. These results indicate that selenium is required by the guinea pig and that a biochemical deficiency state can be induced by dietary means. They also demonstrate the difficulties encountered in assessing selenium status in guinea pigs by measurement of glutathione peroxidase activity. Similar difficulties may be expected in other species.