We investigated the relationship between urinary excretion of MHPG and the clinical response of 17 depressed patients to nortriptyline hydrochloride. Plasma concentrations of nortriptyline were monitored to assure optimal doses. Patients were classified as having "low" or "normal-high" excretion of MHPG based on one to five 24-hour urine specimens. Hamilton Depression Rating Scale scores were not reduced significantly more among the nine low excreters as compared with the eight normal-high excreters. However, when a true bimodal distribution of MHPG excretion was created by comparing only the six lowest excreters with the six highest excreters, the low group improved significantly more than the high group. This differential response to nortriptyline somewhat supports the notion that MHPG excretion may predict response to specific tricyclics. Collecting urine for MHPG determination in depressed patients is not easy; the variability of excretion within patients is considerable, and the range of MHPG excretion closely parallels that in normal persons. The clinical utility of this procedure is still to be determined.