We have investigated the effects of modifying the dose of thyroxine on resting energy expenditure (REE) and on the thermic effect of glucose (TEG) in 9 randomly recruited patients on chronic treatment with this hormone. The initial dose was changed twice in each patient at 6-8 wk intervals, aiming to have a normal, a slightly reduced, and a slightly elevated serum TSH concentration. A total of 27 dose points for each measured variable (3 per patient) were gathered. Dose changes were monitored with serum free T4, T3, and TSH. At the end of each dose period, low density lipoprotein and high density lipoprotein cholesterol, triglycerides, angiotensin converting enzyme, and sex hormone binding globulin were also measured, along with a systematic assessment of symptoms and signs. The investigators involved in the measurements were blinded to the dose of T4. Serum free T4 and TSH significantly correlated to the dose in each patient and in the whole group, whereas serum T3 levels were minimally affected by the dose and did not correlate with it, with free T4 or with TSH. This latter was below normal on 9 occasions, normal in 12, and above normal in 6. Serum free T4 and T3 remained within the normal range on all except 2 occasions. REE and TEG were normalized to fat-free mass (FFM). In each patient there was a significant negative correlation between REE and TSH. This correlation was maintained when all data were pooled (r2 = 0.64; P < 0.001). Also, initial REE and its change between the highest and the lowest thyroxine dose were significantly correlated with, respectively, initial serum TSH (r2 = 0.85; P < 0.001) and the change in serum TSH between the highest and the lowest dose of T4 (r2 = 0.67; P < 0.0065). REE decreased approximately 15% when TSH increased between 0.1 and 10 mU/L. In 6 of the 9 patients, TEG increased with the reduction of the dose, and higher values were associated with higher TSH levels but without reaching statistical significance (F = 2.852, P = 0.077). None of the other indices were significantly affected by the changes in dose. These results indicate that, in patients on chronic treatment with thyroxine, REE is significantly influenced by the dose of this hormone in a dose range encompassing serum TSH concentrations that are considered acceptable in the management of hypothyroid patients. In the absence of physiological or behavioral compensations, these changes in REE may be clinically relevant.