Maximum nocturnal serum melatonin level (MTmax) in relation to some clinical variables was studied in 32 patients with a major depressive episode and in 33 healthy subjects with reference to the outcome of the dexamethasone suppression test (DST). Significant regressions were found between MTmax levels and clinical rating scores in CPRS, interpreted as retardation symptoms. Four healthy subjects with disposition for dysthymic reactions had subnormal MTmax levels, which differed from MTmax levels in subjects without such disposition. Patients but not the healthy subjects, who reported parental loss before 17 years of age, had subnormal MTmax levels and differed from patients with no reported parental loss. Patients with no reported suicidal behaviour in clinical history had significantly lower MTmax levels than patients with reported suicide attempts. No relations were found between low MTmax levels and diagnoses, duration of illness, reported inheritance for depressive illness or sleep disturbances. A hypothetical low melatonin syndrome in depression is proposed: low nocturnal melatonin, abnormal dexamethasone suppression test, disturbed 24-h rhythm of cortisol, less pronounced daily and annual cyclic variation in depressive symptomatology.