Context: For patients on T4 replacement, the dose is guided by serum TSH concentrations, but some patients request higher doses due to adverse symptoms. Objective: The aim of the study was to determine the safety of patients having a low but not suppressed serum TSH when receiving long-term T4 replacement. Design: We conducted an observational cohort study, using data linkage from regional datasets between 1993 and 2001. Setting: A population-based study of all patients in Tayside, Scotland, was performed. Patients: All patients taking T4 replacement therapy (n = 17,684) were included. Main Outcome Measures: Fatal and nonfatal endpoints were considered for cardiovascular disease, dysrhythmias, and fractures. Patients were categorized as having a suppressed TSH (≤0.03 mU/liter), low TSH (0.04-0.4 mU/liter), normal TSH (0.4-4.0 mU/liter), or raised TSH (>4.0 mU/liter). Results: Cardiovascular disease, dysrhythmias, and fractures were increased in patients with a high TSH: adjusted hazards ratio, 1.95 (1.73-2.21), 1.80 (1.33-2.44), and 1.83 (1.41-2.37), respectively; and patients with a suppressed TSH: 1.37 (1.17-1.60), 1.6 (1.10 -2.33), and 2.02 (1.55-2.62), respectively, when compared to patients with a TSH in the laboratory reference range. Patients with a low TSH did not have an increased risk of any of these outcomes [hazards ratio: 1.1 (0.99 -1.123), 1.13 (0.88 -1.47), and 1.13 (0.92-1.39), respectively]. Conclusions: Patients with a high or suppressed TSH had an increased risk of cardiovascular disease, dysrhythmias, and fractures, but patients with a low but unsuppressed TSH did not. It may be safe for patients treated with T4 to have a low but not suppressed serum TSH concentration.