Halofenate, a serum lipid-lowering agent which inhibits binding of thyroid hormone to thyroxine-binding globulin (TBG), was administered daily for 14 days to 8 hypothyroid subjects with elevated TSH concentrations as a result of incomplete thyroxine (T4) therapy. Drug administration resulted in mean increases in serum dialyzable fraction T4 (DFT4) of 52% over pretreatment levels (P less than 0.01) and in dialyzable fraction triiodothyronine (DFT3) of 26% in 7 subjects, (P less than 0.01). During halofenate treatment in these 7 subjects, serum TSH concentrations decreased significantly (mean = 39%, P less than 0.01) when DFT4 and DFT3 were increased by halofenate. In only two subjects was there a convincing temporal relationship between increased serum absolute free T4 (AFT4) and decreased serum TSH concentrations. Contrary to what would be predicted from the "free hormone hypothesis", changes in serum TSH concentration in these hypothyroid patients appeared to relate primarily to changes in the free fraction of circulating T4 and T3 (DFT4, DFT3), rather than to alterations in AFT4 or AFT3. Halofenate did not alter serum TBG binding capacity. An eighth subject did not show increased DFT4 and DFT3 during halofenate treatment despite achievement of therapeutic serum levels of the agent; in this patient, serum TSH levels rose progressively throughout the period of inadequate T4 replacement and halofenate administration. In hypothyroid patients, short-term halofenate use suggests that the pituitary-thyroid hormone feedback circuit can respond to increases in serum DFT4 and DFT3 in the absence of detactable increases in absolute free hormone concentrations.