Over 1% of the UK population is receiving thyroid hormone replacement with l-thyroxine (T4). However, many patients complain of persistent lethargy and related symptoms on T4 even with normal TSH levels. To date there has been no large study to determine whether this is related to thyroxine replacement or coincidental psychological morbidity. We have therefore attempted to address this issue using a large, community-based study. Computerized prescribing records of five general practices were used to identify 961 patients who had been on thyroxine for a minimum of 4 months from a population of 63 000 (1.5%), along with age- and sex-matched controls. All 1922 individuals were sent a two-page questionnaire, made up of the short form of the General Health Questionnaire (GHQ-12), designed to detect minor psychiatric disorders in the community, and a 12-question 'thyroid symptom questionnaire' (TSQ) in the same format. A covering letter explained that we were interested in 'how patients felt on medication' and made no direct reference to thyroxine. Scores from the GHQ and TSQ were marked for each individual using the GHQ and Likert scoring methods. Patients' latest TSH measurements were obtained from laboratory records. Comparisons were then made on scores for the total GHQ-12, TSQ and individual questions between patients (P) and control (C) groups. Separate analyses were made comparing the patients with a normal TSH (nP) and the control group. Five hundred and ninety-seven (62%) of the patients (P) and 551 (57%) of the controls (C) responded and fully completed at least one of the two questionnaire. Three hundred and ninety-seven responding patients (nP) had a TSH estimation performed in the previous 12 months with the last result being in the local laboratory normal range for TSH (0.1-5.5 or 0.2-6.0 mU/l, according to the assay method used). The responding P, nP and C populations were well matched for age (59.96, 59.73, 59.35 years) and sex (85%, 83%, 87% female). The number of individuals scoring 3 or more on the GHQ-12 (indicating 'caseness') was 21% higher in P than C [185/572 (32.3%) vs. 137/535 (25.6%), P = 0.014] and 26% higher in nP than C [131/381 (34.4%) vs. 137/535 (25.6%), P < 0.005]. Stronger differences were seen with the TSQ scores [C = 187/535 (35.0%), P = 273/583 (46.8%), P < 0.001, P vs. C; and nP = 189/381 (48.6%), P < 0.001, nP vs. C]. Differences existed in chronic drug use and chronic disease prevalence between the control and patient groups, but the differences in GHQ and TSQ scores between the groups remained significant even after correction for these factors as well as age and sex in multiple regression analysis. This community-based study is the first evidence to indicate that patients on thyroxine replacement even with a normal TSH display significant impairment in psychological well-being compared to controls of similar age and sex. In view of the large numbers of people on thyroxine replacement, we believe that these differences, although not large, could contribute to significant psychological morbidity in a substantial number of individuals.