Objective: To report a series of newly diagnosed thyrotoxic patients with concurrent acute psychosis, and to assess the association between the two disorders. Design: Retrospective study of thyrotoxic patients with associated psychosis ('thyrotoxic psychosis'; TP) requiring inpatient psychiatric care. New Zealand thyrotoxicosis annual incidence figures and first psychiatric admission rates for affective psychosis were utilised to statistically assess the co-occurrence of thyrotoxicosis and affective psychosis. Patients and Methods: During the 20-year study period, 18 inpatients (16 women and 2 men), mean age 54 years, with TP were identified. No patient had a past history of thyrotoxicosis, but four had required psychiatric inpatient care many years earlier. Thyrotoxicosis was documented by radioimmunoassay of thyroid hormone levels, and thyroid scintiscan. Psychiatric manifestations were classified using ICD9 criteria. Results: Thyroid hormone levels were markedly elevated in more than half of our TP patients. All younger patients had Graves' disease, and most older patients toxic nodular goitre. All patients were treated with antithyroid drugs, and all but one subsequently received 131 I therapy. Two patients were not mentally ill when thyrotoxicosis was diagnosed, but suffered major mood swings when thyroid hormone levels were falling. There was no specific psychiatric clinical picture but affective psychoses were commonest -seven depression, seven mania. The other diagnoses were two schizophreniform, one paranoid, and one delirium. Initially, neuroleptic medication was used in all but one patient, and during long-term follow-up (median 11 years) more than half our series had remained well with no further psychiatric problems. Statistical analysis was restricted to thyrotoxic patients with first psychiatric hospital admission for affective psychosis. During the 20-year period, there were nine thyrotoxic patients (95% confidence interval 4.5-17.1) with concurrent affective psychosis requiring first admission, and the calculated expected number was only 0.36. These findings indicate a clear association well above chance co-occurrence. Conclusion: TP is not a specific clinical picture, but affective psychoses are commonest. Statistical analysis of thyrotoxic patients with concurrent affective psychoses showed an incidence well above chance co-occurrence. It appears that thyrotoxicosis may be a precipitant of acute affective psychosis.