Abstract Among previous studies on monoamine metabolites in human cerebrospinal fluid (CSF) of depressed patients and normal controls, inconsistent results have been observed and those may have been partly caused by the confounding background factors. The aims of the present study are following; first, relatively large sample size were used; second, only patients with major depressive disorder (MDD) (not including bipolar disorder) were included; third, background variables such as sex and age were strictly controlled; lastly, diagnosis, medication and severity of depression were simultaneously analyzed on CSF monoamine metabolite levels. Human subjects, including healthy control, were recruited from the outpatient clinic of the National Center Hospital, National Center of Neurology and Psychiatry (Tokyo, Japan), or through advertisement in magazines and our homepages. CSF was collected from 75 patients with MDD and 87 healthy controls. As previous reports, the strong correlation between HVA and 5-HIAA was observed in CSF of our healthy subjects and its correlation was weaker in CSF with MDD patients. Although HVA level did not show any difference between healthy controls and MDD patients, when MDD group were divided into the severity of depression, it was significantly decreased in CSF of MDD patients with high points (HIGH; more than 13 points) of GRID Hamilton Depression Rating Scale, 17-item version (HAMD-17) score. Antidepressant treatments did not affect the HVA reduction. 5-HIAA level was also decreased in HIGH group. Unlike HVA, 5-HIAA level was affected by both antidepressant and severity of MDD. 3-methoxy-4-hydroxy-phenylethyleneglycol (MHPG; noradrenaline metabolite) level was only decreased by antidepressant treatments. Based on these results, we reinterpreted the monoamine hypothesis in MDD and these results suggest some clues for the next generation of anti-depressant.