Twenty psychiatric patients on lithium medication were examined with 7-Li-magnetic resonance spectroscopy of the brain. Patients on long-term lithium treatment (> 6 months) were compared with a short-term group who had been taking lithium for between 4 and 8 weeks. Patients met DSM-III-R criteria for either recurrent unipolar depressive disorder (DSM-III-R 296.3x) or schizoaffective disorder, depressive type (DSM-III-R 295.70). The brain:serum lithium ratio was 0.76 +/- 0.26; there was no significant difference between short-term and long-term treatment. In the group of long-term treatment patients there was a positive correlation between lithium dose per day and brain lithium concentration (R = .72, p < .01), and between lithium plasma concentration and brain lithium concentration (R = .65, p < .05). In the short-term group, however, there was no significant correlation for these parameters. No differences between unipolar and schizoaffective disorder were found.