Twenty-five women with unipolar primary major depressive disorder (20 endogenous, 5 nonendogenous) and 20 female control subjects were studied with the thyrotropin-releasing hormone stimulation test (TRH-ST). Prolactin (PRL) levels were measured before and after TRH administration for patients and control subjects. For patients, thyrotropin (thyroid-stimulating hormone; TSH) levels were measured from the same serum specimens as PRL levels, and the 1 mg dexamethasone suppression test (DST) was performed. Patients with endogenous depression (ED) had significantly lower maximal serum PRL levels (max PRL) following TRH, and a significantly reduced increase over basal serum PRL (delta max PRL) compared to normal controls (NC). Nonendogenous depressed (NED) patients did not differ significantly from the ED or NC groups on either of these measures. For the ED group, delta max PRL was inversely correlated with severity of depressive symptomatology. Basal PRL levels did not differentiate the depressed subgroups (ED, NED) from each other or from the NC group. Depressed patients with blunted delta max PRL values tended to have blunted delta max TSH values and vice versa. Almost all patients with blunting of either delta max PRL or delta max TSH were also DST nonsuppressors; conversely, only about half (7 of 12) of patients who were DST nonsuppressors had either blunted delta max PRL or delta max TSH. Patients with clinical diagnoses of melancholia or psychotic features were significantly more likely to have blunted delta max PRL values than patients without these diagnoses. A disturbance of central noradrenergic function could explain these findings.