Sixty-nine depressive patients (DSM III criteria: 296.2, 296.3, 296.5, 300.4) were treated with 40 to 60 mg citalopram (CIT) daily for 4 weeks. Among them, 45 responded to treatment (improvement > 50% on the 21-item Hamilton Rating Scale for Depression [HAM-D]) and continued their treatment for another week before being released from the study. The 24 nonresponders were randomized and comedicated under double-blind conditions with lithium carbonate (Li) (2 x 400 mg/day) (CIT-Li group) or with placebo (CIT-Pl group) from days 29 to 35. For days 36 to 42, the patients of both subgroups were treated openly with Li (800 mg/day) in addition to the ongoing CIT treatment. On day 35, 6 of 10 patients responded to the CIT-Li combination, whereas 2 of 14 patients only responded to the CIT-Pl combination. This group difference reached significance (p < 0.05) on day 35 with lower HAM-D total scores in the CIT-Li group. No evidence was seen of a pharmacokinetic interaction between CIT and Li, and this combination was well tolerated. Patients were phenotyped with dextromethorphan and mephenytoin at baseline and at day 28. As evaluated at baseline, three patients (responders) were poor metabolizers of dextromethorphan and six patients (three responders and three nonresponders) of mephenytoin. On day 28, the ratio CIT/N-desmethylCIT (DCIT) in plasma was significantly higher in poor than in extensive metabolizers of mephenytoin (p = 0.0001), and there was a significant positive correlation between the metabolic ratio of dextromethorphan and the ratio DCIT/N-didesmethylCIT in plasma (p < 0.001). These findings illustrate the role of CYP2D6 and CYP2C19 in the metabolism of CIT. It can be concluded that Li addition to CIT is effective in patients not responding to CIT alone without any evidence of an accentuation or provocation of adverse events.