OBJECTIVE Cognitive and affective functioning is sensitive to changes in thyroid hormones. We have sought to determine: (1) the prevalence of thyroid function abnormalities in a psychiatric population on admission (as compared to the prevalence in a normal population), and (2) whether such thyroid function abnormalities are associated with the occurrence or development of cognitive and affective disorders.DESIGN Serum was collected 2–3 weeks after hospitalization in 3 major clinics from 3756 psychiatric patients in 1987–1990, stored, and assayed in 1993 for the presence of antibodies against the TSH‐receptor and thyroperoxidase (TPO‐Ab) and for TSH levels. The psychiatric cohort was matched with a control population of healthy individuals living in the same area (n = 1877).The prevalence study was followed by a case‐control study involving patients from one clinic that had routinely assigned a DSM‐IIIR classification to its patients. Cases were those admissions with thyroid abnormalities and three subgroups of cases were randomly formed demonstrating either TSH less than 0.4 mU/l (n = 44) or over 4.0 mU/l (n = 44), or TPO‐Ab positivity (n = 50). Cases were compared to random controls from the same psychiatric population, viz patients without thyroid abnormalities (n = 83). Comparison was with respect to their psychiatric follow‐up diagnosis (the investigator was blinded to the thyroid test results).RESULTS Prevalence study. The percentage of patients positive for TSH‐receptor‐Ab was 0.26 (9/3504), for TPO‐Ab was 10.0 (331/3316) and outside the TSH range of 0.4–4.0 mU/l was 10.0 ((332/3316): 5.9% (198/3316)>4.0 mU/l and 4.1% (134/3316)<0.4 mU/l). Abnormal total thyroxine levels were found in only 9.8% of subjects with abnormal TSH, indicating the predominantly subclinical character of the thyroid alteration. In comparison, the healthy area controls over 55 years of age showed the same prevalence of positive TPO‐antibodies and TSH under 0.4 mU/l, but a higher prevalence of TSH over 4.0 mU/l.Case‐control study. In the case control analysis differences could not be noticed with regard to prevalences of dementia, schizophrenia or other psychiatric illnesses apart from the prevalence of affective disorders which were more prevalent in TPO‐Ab positive patients and patients with a low serum TSH. Since prior use of lithium, carbamezapine, carbimazole and/or thyroxine could be a factor of importance in this association, analyses were also carried out excluding patients with such prior drug use. In these analyses affective disorders were still more prevalent in patients with a low serum TSH (particularly in males, 40% in casesvs9% in controls,P  4.0 mU/l) and a subgroup of the affective disorders, viz with a rapid cycling of bipolar disorder (18% in casesvs0% in controls,P < 0.001).CONCLUSION Thoug h causal relations cannot be determined from this cross‐sectional study, this admission survey found early forms of autoimmune thyroid disease, sometimes characterized only by TPO‐Abs, highly significantly associated with rapid cycles of a bipolar disorder. It also found a weak association between s