A 2-site immunoradiometric assay for serum ferritin was evaluated with commercially available kit. The assay required 6 hours. The slope of the standard curve kept up ideal range with the calculation of maxium binding instead of total dose until expire date. The stage II washing was more important than the stage I washing on the modified washing procedure as the bead keeping to remain in the tube. With this modified mothod, three times of tube. washing was sufficient to reduce the significant errors The measured values of serially diluted sample with standard diluting buffer was proportional to the predicted values. In the experiment of serum effect on the assay. a linear relationship from 5 to 50% serum, but beyond 50% there was reduction in measured ferritin concentration. It has a sensitivity of 2.77 ng/ml, within-assay precision (CV) of 8.0%, and between-assay reproducibility(CV) of 7.4% (mean 174.8 ng/ml).