The purpose of this study was to investigate the effects of milnacipran and paroxetine on plasma levels of catecholamine metabolites, and we attempted to elucidate the differences between the mechanisms of these drugs in catecholaminergic neurons. In depressed patients, we investigated the relationships among pretreatment levels of catecholamine metabolites, the changes in plasma catecholamine metabolite levels before and after administration of milnacipran or paroxetine, and clinical response to these drugs. Responders to milnacipran showed lower pretreatment levels of plasma 3-methoxy-4-hydroxyphenylglycol (pMHPG) than did nonresponders to milnacipran; there was also a positive correlation between changes in pMHPG levels and percent improvement of the score on the 17-item Hamilton Rating Scale for Depression (HRSD). On the other hand, responders to paroxetine showed higher pretreatment levels of pMHPG than did nonresponders to paroxetine, and a negative correlation was observed between changes in pMHPG levels and percent improvement of the HRSD score. However, a significant difference was not observed in the pretreatment plasma level of homovanillic acid between responders and nonresponders to treatment with milnacipran or paroxetine. These results suggest that there is an association between baseline pMHPG levels and clinical responses with respect to milnacipran versus paroxetine treatment.