The mouse spleen cell assay (MSCA) has been compared with a radioimmunoassay for the measurement of serum erythropoietin (Ep). In 20 normal subjects the serum values ranged from 15 to 73 mU/ml for the MSCA compared with 5-30 mU/ml for the RIA. For normal sera there was no correlation between the results of the two assays. In 37 patients with anaemias of differing aetiologies and at various stages of treatment values ranged from 10 to 3645 mU/ml for the MSCA and 13-10,000 mU/ml for the RIA. Although patient values from the two assays were highly correlated (r = 0.98, P less than 0.001), the MSCA results were generally lower. These discrepancies can be largely accounted for by two factors. Firstly the MSCA is sensitive to non-specific matrix effects. Secondly, heat inactivation of serum, a prerequisite for the MSCA, but not for the RIA, destroys a variable and unpredictable proportion of the Ep in the test sera leading to an underestimation of Ep in the MSCA. We conclude that the RIA is more reliable than the MSCA which, in its present form, cannot be recommended for the accurate measurement of serum erythropoietin.