Gel filtration of human RBC lysate incubated with labeled T4 or T3 revealed co-elution of a major iodothyronine-binding fraction (R-2) and hemoglobin. Solutions of purified human hemoglobin and T3 also showed co-elution of hormone and hemoglobin. Because hematin and protoporphyrin were shown to bind labeled T3, the oxygen-binding site on hemoglobin was excluded as the site of iodothyronine-hemoglobin interaction. Analysis of hormone binding by heme and globin moieties showed T3 binding to be limited to the heme fraction. Addition of excess unlabeled T3 to hemoglobin or heme incubated with labeled T3 indicated 75% to 90% of hormone binding was poorly dissociable. These observations suggested that the presence of hemoglobin in RBC lysate or in serum could influence the measurement of T4 and T3 by specific RIA. Subsequent studies of the addition to serum of human hemoglobin revealed a significant reduction in T3 and T4 detectable by RIA in the presence of this protein. The effect was influenced by the concentration of hemoglobin and by duration and temperature of incubations of hemoglobin and serum prior to RIA. Incubated for 5 days at 4 degrees C, 14 sera containing 10 gm/dl hemoglobin showed a mean decrease in T3 concentration of 40% compared to sera incubated in the absence of hemoglobin (160.1 to 93.9 ng/dl, p less than 0.001); detectable serum T4 fell by 50% in 13 sera incubated under the same conditions (5.40 micrograms/dl without hemoglobin to 2.55 micrograms/dl in the presence of hemoglobin, p less than 0.001). Hemoglobin concentrations in serum as low as 0.1 and 0.5 gm/dl affected the RIAs significantly. Thus a major fraction of thyroid hormone binding in human RBC cytoplasm is accounted for by an interaction with hemoglobin. This interaction in serum or RBC lysates is a significant variable affecting iodothyronine determinations.