The accurate measurement of biologically active erythropoietin (Ep) in human serum and plasma using present in vivo and in vitro bioassays is difficult because of the presence of both inhibitors and non-Ep stimulators of erythropoiesis. We have developed a simple procedure to quantitatively purify Ep from serum and plasma for subsequent testing in the phenylhydrazine-treated mouse spleen cell assay. The method involves absorption of Ep to an immobilized high-affinity anti-Ep monoclonal antibody and acid elution of the antibody-bound material. After neutralization, the eluted EP is then tested directly in the in vitro bioassay without interference by other serum proteins. By using magnetic beads as a solid support for the antibody, washing and elution steps can be performed rapidly and efficiently. Recoveries of Ep after this procedure show very little sample-to-sample variation and are consistently between 45% and 55%, which is close to the maximum binding expected for the anti-Ep antibody. Coupled with the 7.4-fold concentration that this procedure affords, there is an overall increase in sensitivity of three- to fourfold, which makes this assay suitable for accurately measuring Ep levels in patients with below-average titers. Results with this magnetic bead assay indicate that accurate and reproducible estimates for Ep levels in the serum and plasma from healthy donors as well as from patients with hematologic disorders can be obtained. Titers of biologically active Ep in the sera from a group of patients with either leukemia or lymphoma were found to be elevated, and the values correlated well with titers of immunoreactive Ep measured in the Ep radioimmunoassay. Because of its specificity and high sensitivity, the magnetic bead assay is a valuable alternative to immunoassays for the measurement of elevated, normal, and even subnormal Ep levels in human serum and plasma.