A new method for the estimation of the bioavailability of thyroxine (T4) and 3,5,3'-triiodothyronine (T3) is described based on gel separation followed by antibody extraction of labelled T4 and T3 from serum, and using the area under the curve of disappearance of the tracer (AUC) for the calculations. The peak serum concentrations of radioactive labelled T4 and T3 were reached approximately 90 min after oral administration of both tracers. The relative difference of duplicate estimations was below 10% (n = 3). The bioavailability of T4 in 6 euthyroid controls was in median 65% (range 64-75%), and it was significantly increased both in hyperthyroidism (88% (75-99%), n = 6, P less than 0.01) and hypothyroidism (84% (67-100%), n = 6, P less than 0.02). The bioavailability of T3 in 6 euthyroid controls was in median 78% (69-99%) and significantly greater than that of T4 (P less than 0.02). The bioavailability was unaffected by hyperthyroidism (79% (61-98%), n = 9) and hypothyroidism (77% (66-97%), n = 7). No significant difference between T4 and T3 bioavailabilities was found in hyper- or hypothyroidism. The clinical implication of the present study is that the bioavailability of T4 and T3 is almost identical and approximately 80% in patients with severe hypothyroidism.