Studies that aim at identifying genes or environmental factors contributing to the development of autoimmune thyroid disease (AITD) demand that several hundred patients and control subjects be assessed. In these large studies, the laboratory methodology is often described in detail whereas little attention is given to an accurate description of the study population. Usually, a diagnosis of AITD in the control group is based on self-reported disease status. Although such studies have been criticized for diagnostic inaccuracy, no study has evaluated the validity of self-reported hyperthyroidism and hypothyroidism in detail. We have assessed the validity of self-reported hyperthyroidism and hypothyroidism in 401 twin pairs from among 6,628 same gender pairs, ages 18-41 years who participated in a nationwide questionnaire survey in 1994. The self-reported questionnaire data were compared with information from medical records and the 1/kappa coefficient, sensitivity, and specificity were determined. Overall, there was only a slight to fair agreement between the self-reported questionnaire data and medical record data as shown by kappa values of 0.18, 0.21, and 0.26 for hyperthyroidism and hypothyroidism as a group, hyperthyroidism and hypothyroidism, respectively. For both hyperthyroidism and hypothyroidism, the sensitivity of the self-reported diagnosis was 0.98, whereas the specificity was 0.57 and 0.67 for self-reported hyperthyroidism and hypothyroidism, respectively. In conclusion, the validity of self-reported hyperthyroidism and hypothyroidism is unsatisfactorily low. However, by combining self-reports with valid retrospective data on diagnostic findings it can be used as a sampling method in large epidemiological or genetic studies.