Some characteristics of a radioimmunoassay of selenoprotein P (a major selenoprotein in human serum) are described. Polyclonal antibodies generated in rabbits were used and goat anti-rabbit-IgG antiserum was used as a second antibody. Depending on the concentration of selenoprotein P, 1–10 µl of human serum were used in the assay. The relative standard deviation for the concentration of selenoprotein P was 6.3% between assays and 7.7% within assays. Different animal sera gave no significant interference, indicating that the antibodies did not react with non-human analogues of selenoprotein P. No indication of cross-reactivity could be found concerning extracellular glutathione peroxidase (another selenoprotein in serum). Addition of increasing amounts of normal human serum and partially purified selenoprotein P to the radioimmunoassay resulted in parallel curves. Incubation at 4 °C gave somewhat higher binding of labelled selenoprotein P than incubation at room temperature. The epitope, recognized by the antibodies, was apparently stable after storage of serum (in the frozen state for years, and in the cold for months). No significant amount of selenoprotein P could be demonstrated in red blood cells, and analysis of haemolysed whole blood gave expected data. Investigations of selenium status in different study groups indicated that in most cases the concentration of selenoprotein P in serum was positively correlated to that of glutathione peroxidase and serum selenium. In an intervention study, where subjects decreased their selenium intake to 50%, the serum levels of glutathione peroxidase and selenium decreased, but no significant decrease of selenoprotein P could be demonstrated. In conclusion, the data suggest that selenoprotein P may be a valuable new biochemical marker of selenium status, which may respond differently from selenium and glutathione peroxidase under certain conditions.