The finding of a diminished TSH response to exogenously administered TRH in a significant proportion of depressed patients has now been established as one of the most reproducible findings in biological psychiatry. More than 50 reports, in which more than 1000 patients have been studied, reveal that the TSH response is blunted in approximately 25% of patients with major depression. TSH blunting is clearly not specific for depression, because it also has been observed in mania, alcoholism, and borderline personality disorder. It is doubtful that TSH blunting represents a non-specific response to mental stress: it was found only rarely in schizophrenic patients, and the TSH response to in vivo flooding therapy in phobic patients was normal. In both depression and alcoholism, TSH blunting has been reported to be sometimes a state marker and sometimes a trait marker, i.e. the fault was found to persist into remission in more than half the patients. In both conditions, TSH blunting was unrelated to the patients' age, body weight, height, body surface, thyroid status, and serum cortisol concentrations. It also is unlikely that TSH blunting was due to increased dopaminergic inhibition of thyrotroph cells: serum prolactin concentrations in TSH blunters were found to be normal, and pretreatment with haloperidol had no effect on either basal TSH levels or TSH blunting. In depression, TSH blunting was not associated with previous drug intake, dexamethasone suppression test abnormalities, or variables of biogenic amine metabolism, nor did it usefully segregate between primary and secondary depression or between unipolar and bipolar subgroups. Preliminary evidence suggests that TSH blunting in depression may be related to duration of illness, history of violent suicide attempts, and a reduced 24 h TSH secretion. In alcoholism, TSH blunting was unrelated to family or personal history of depression and duration of abstinence. With reference to clinical utility, TSH blunting may aid in assessing the response to antidepressant treatment, predicting outcome to such treatment, assessing the risk for violent suicide attempts, and describing relationships between different psychiatric populations (e.g. depression and alcoholism).