Subclinical hypothyroidism is a frequent syndrome affecting about 10 million people in the United States. The management of such patients is open to debate. In a long-term prospective study we analyzed the spontaneous course and the value of predictive factors in the development of overt thyroid failure. We studied 82 female patients with subclinical hypothyroidism prospectively over a mean observation period of 9.2 yr. TSH, thyroid hormones, thyroid reserve after TRH administration, thyroid antibodies, and clinical parameters were assessed at yearly intervals. The cumulative incidence of overt hypothyroidism was calculated using life-table analysis and Kaplan-Meier curves. According to the initial serum TSH concentrations (TSH, 4-6/>6-12/>12 mU/liter), Kaplan-Meier estimates of the incidence of overt hypothyroidism were 0%, 42.8%, and 76.9%, respectively, after 10 yr (P < 0.0001). When only patients with TSH levels greater than 6 mU/liter were analyzed, the cumulative incidence was 55.3%. The incidence of overt hypothyroidism increased in patients with impaired thyroid reserve (52.6% vs. 38.1%; P = 0.05) and positive microsomal antibodies (58.5% vs. 23.2%; P = 0.03). This prospective long-term study demonstrates that only a part of the cohort of patients with subclinical hypothyroidism develops overt hypothyroidism over time and that a major group remains in the subclinical state after 10 yr. The measurement of TSH, microsomal (thyroperoxidase) antibodies, and thyroid reserve allows initial risk stratification for the development of overt thyroid failure (risk ratio ranging from 1.0-15.6). Our study helps to recognize the spontaneous course of subclinical hypothyroidism and in the identification of patients most likely to progress to overt hypothyroidism.