Twenty-four hour urinary excretion of 3-methoxy-4-hydroxphaeylglycol (MHPG), the metabolite thought best to reflect brain norepinephrine metabolism, was studied longitudinally in ten depressed patients before and during the acute and chronic phases of lithium treatment. Five of the patients were identified as bipolar I (prior history of mania), 3 as bipolar II (history of hypomania) and 2 as unipolar (history of depression). During acute lithium administration (first week) there was no consistent pattern of change in MHPG. Comparing the predrug period with the third and fourth week of treatment, all of the responders showed an increase in MHPG, while the non-responders showed no change or a decrease. It is concluded that the change in clinical state is the most important variable contributing to MHPG changes in these patients. There was a tendency for the pretreatment MHPG excretion to be low in the patients who went on to show a clear-cut antidepressant response to lithium compared to those who were unequivocal non-responders. The predrug MHPG for the bipolar patients (prior history of mania) was significantly lower than the unipolar patients, a difference which apparently contributes to the lower MHPG in the lithium responders, all of whom were in the bipolar group.