We investigate whether the prognosis of patients with differentiated thyroid cancer is improved by maintaining a greater level of TSH suppression. One hundred and forty-one patients who underwent hormone therapy after thyroidectomy were followed up from 1970 to 1993 (mean, 95 months). Patients received levothyroxine (L-T4; mean dose, 2.6 micrograms/kg-day). TSH suppression was evaluated by TRH stimulation test until 1986 and thereafter by a second generation immunoradiometric assay. As TSH underwent fluctuation over time in most patients, we focused on subgroups of patients with relatively constant TSH levels during the follow-up. The relapse-free survival (RFS) was longer in the group with constantly suppressed TSH (all TSH values, < or = 0.05 mU/L; n = 18) than in the group with nonsuppressed TSH (all TSH values, > or = 1 mU/L; n = 15; P < 0.01). Age, sex, tumor node metastasis stage, and initial therapy were not different between the suppressed and nonsuppressed TSH groups. In the overall population, we analyzed the level of TSH suppression by studying the percentage of undetectable TSH values (< or = 0.05 mU/L) during the follow-up. The patients with a greater degree of TSH suppression (> 90% of undetectable TSH values; n = 19) had a trend toward a longer RFS than the remaining population (n = 102; P = 0.14). The patients with a lesser degree of TSH suppression (< 10% of undetectable TSH values; n = 27) had a shorter RFS than the remaining patients (n = 94; P < 0.01). In multivariate analysis that included TSH suppression, age, sex, histology, and tumor node metastasis stage, the degree of TSH suppression predicted RFS independently of other factors (P = 0.02). This study shows that a lesser degree of TSH suppression is associated with an increased incidence of relapse, supporting the hypothesis that a high level of TSH suppression is required for the endocrine management of thyroid cancer.