Use of sensitive assays for TSH in the follow-up of patients treated with radioiodine (131I) for thyrotoxicosis has led to questions regarding the significance of abnormal TSH values found in association with normal circulating thyroid hormone levels, especially TSH results below normal. We have investigated the relationship between serum TSH and the likelihood of maintenance of euthyroidism, as well as the relationship between serum TSH and free T4, in 389 subjects treated with 131I 2-35 years previously who had free T4 and free T3 values within the normal range and who were not receiving thyroxine or antithyroid therapy. In those with undetectable TSH (less than 0.05 mU/L), TSH remained undetectable in 54.5% at 1 yr but rose to detectable or normal values in the remainder. In those with low but measurable TSH (0.05-0.5 mU/L), results were similar at 1 yr in 47.5% and returned to normal in 45%. No patient with a TSH value below normal became hypothyroid (defined as a reduction in serum free T4). In those with normal TSH (0.5-5.0 mU/L) at time 0, TSH remained normal in 83%, fell in 4% and became elevated in 13%. The yr 1 incidence of hypothyroidism was 1%; one patient became thyrotoxic. In those with TSH values above normal (5.0-15.0 mU/L), TSH remained elevated in 90.6%; the incidence of hypothyroidism was 14.5% in yr 1. The small risk of development of hypothyroidism in those with subnormal or normal TSH indicates that biochemical testing is not essential for 1-2 yr in such patients; this contrasts with a need for repeat testing within a period not exceeding 1 yr in those with elevated TSH. The relationship between serum-free T4 and TSH suggests that TSH results outside the normal range reflect thyroid hormone excess or deficiency. Persistence of undetectable TSH values during follow-up and the observation of higher free T4 at time 0 in those whose TSH remained undetectable compared with those whose TSH rose suggest that undetectable TSH concentrations are of greater significance than low but detectable values.