The purpose of this study was to compare the overall effectiveness of 2 treatment strategies for inpatients with severe major depressive episode (DSM-III-R): (1) mirtazapine (phase 1) and subsequent lithium addition (phase 2) or (2) imipramine (phase 1) and subsequent lithium addition (phase 2). We previously reported the results of phase 1. In phase 1, patients were randomly assigned to treatment with either mirtazapine or imipramine, and doses were adjusted to obtain predefined blood drug levels. Nonresponders had lithium added to the double-blind mirtazapine or imipramine medication. The dose was adjusted to obtain a blood lithium level of 0.5-1.0 mmol/L. Treatment effects were evaluated weekly by the Montgomery-Asberg Depression Rating Scale for up to 2 weeks on this blood lithium level. Data for 100 patients were available for comparison of the 2 treatment strategies. 80 patients received no comedication. By the end of phase 2, 24 (48%) of 50 had responded to mirtazapine and 32 (64%) of 50 had responded to imipramine (intent-to-treat analysis). A survival analysis of the total patient group intent-to-treat showed a significant difference in favor of the treatment strategy with imipramine and subsequent lithium addition. Efficacy of imipramine and subsequent lithium addition for nonresponders is superior to the same treatment strategy with mirtazapine. This applies to the patient sample studied, which consisted of 100 severely depressed inpatients, 29 of whom were psychotically depressed. More serious side effects of imipramine, however, led to discontinuation of imipramine in 5 patients. No serious side effects were observed during the phase of lithium addition to either imipramine or mirtazapine. We, therefore, prefer to start treatment with imipramine and test for fixed blood drug levels, and, if necessary, add lithium. In the case of prohibitive side effects, patients are switched to a modern antidepressant such as mirtazapine, and, if necessary, lithium is added to this antidepressant.