Two experiments were conducted to establish responses in milk Se concentrations in grazing dairy cows to different amounts of dietary Se yeast, and to determine the effects of the Se concentration of the basal diet. The hypothesis tested was that the response in milk, blood, and tissue Se concentrations to supplemental Se would not be affected by whether the Se was from the basal diet or from Se yeast. In addition, by conducting a similar experiment in either early (spring; experiment 1) or late (autumn; experiment 2) lactation, we hypothesized that different Se input-output relationships would result. Both 6-wk experiments involved 60 multiparous Holstein-Friesian cows, all of which had calved in spring. They were allocated to 1 of 10 dietary Se treatments that included 2 types of crushed triticale grain (low Se, approximately 165 microg of Se/kg of DM; or high Se, approximately 580 microg/kg of DM) fed at 4 kg of DM/d, and 1 kg of DM/d of pellets formulated to carry 5 quantities of Se yeast (0, 4, 8, 12, or 16 mg of Se). Daily total Se intakes ranged from <2 to >18 mg/cow in both experiments. Milk Se concentrations plateaued after 15 and 7 d of supplementation in experiments 1 and 2, respectively, and then remained at plateau concentrations. Average milk Se concentrations for the plateau period increased as the amount of Se yeast increased, and low- and high-Se grain treatments were different at all quantities of Se yeast, although there was a tendency for this difference to diminish at the greatest concentrations of yeast. There were significant positive, linear relationships between Se intake and the concentrations of Se in milk, which were not affected by the source of Se, and the relationships were similar for both experiments. Therefore, the output of Se in milk in experiment 1 was greater than that in experiment 2 because the milk yield of the cows in early lactation was greater. The estimated proportions of Se partitioned to destinations other than milk and feces increased with the amount of Se in the diet and were greater in experiment 2 than in experiment 1, a result that was supported by Se concentrations in whole blood and plasma and in semitendinosus muscle tissue. If high-Se products are to be produced for human nutrition, it is important to be able to develop feeding systems that produce milk with consistent and predictable Se concentrations so that products can consistently meet specifications. The results indicate that this objective is achievable.