We have examined the performance of a commercial free-thyroxine assay in which a radiolabeled T4 derivative permits the competitive quantitation of extracted T4 in the presence of serum proteins. After the total T4 pool had been radiolabeled with either I-125 T4 or I-131 T4, the solid-phase antibody was found to be associated with 4-8% of the total T4 present in the assay tube. Of this, 15-60% was displaceable (antibody-bound). The assay estimated free T4 to be 0.6-1.8 ng/dl in euthyroid patients, and distinguished them from hyperthyroid (sensitivity 91%) and hypothyroid patients (sensitivity 91%) without apparent TBG dependence. In patients with severe nonthyroidal illnesses, the assay correctly quantitated a reduced extracted mass in some. In other patients, however, the assay results were inappropriately lower than the actual extracted mass, in agreement with the FTI but not with the measurements of free T4 by dialysis. This assay appears to produce clinically appropriate results in most patients. In some nonthyroidally ill patients however, the indicated free T4 is spuriously low.