Forty-five patients with primary hypothyroidism were studied during the first 4 to 36 months of replacement therapy with thyroxine. All became clinically euthyroid (23 patients while taking 0.1 mg/d, 14 patients while taking 0.15 mg/d, 7 patients while taking 0.2 mg/d, and 1 patient while taking 0.25 mg/d) over a period of three to six months. The patients were then divided for data analysis into two groups. Group I had normal serum T3 levels, normal (or elevated) serum T4 levels, and normal serum TSH levels. Group II had normal serum T3 levels, normal (or elevated) serum T4 levels, but high serum TSH levels. Group II was subdivided further into a group of 13 patients (group IIa) whose dose of thyroxine was deliberately increased until the serum TSH level was normalized; five of these patients became clinically and biochemically hyperthyroid. Group IIb consisted of eight patients with normal serum T3 and T4 levels and high serum TSH levels who were followed up without attempting to normalize their serum TSH levels. None became thyrotoxic, and their serum TSH levels showed little change. These findings suggest that serum TSH levels alone are not adequate to assess the required dose of thyroxine replacement therapy.