New evidence indicates that treatment response can be predicted with high sensitivity after 2 weeks of treatment. Here, we assess whether early improvement with antidepressant treatment predicts treatment outcome in patients with major depressive disorder (MDD). Forty-one clinical trials comparing mirtazapine with active comparators or placebo in inpatients and outpatients (all-treated population, N = 6907; intent-to-treat population, N = 6562) with MDD (DSM-III-R or DSM-IV Criteria) were examined for early improvement (>or= 20% score reduction from baseline on the 17-item Hamilton Rating Scale for Depression [HAM-D-17] within 2 weeks of treatment) and its relationship to treatment outcome. Data were obtained from a systematic search of single- or double-blind clinical trials (clinical trials database, Organon, a part of Schering-Plough Corporation, Oss, The Netherlands). All included trials (a total of 41) employed antidepressant treatment for more than 4 weeks and a maximum of 8 weeks. The studies ranged from March 1982 to December 2003. Trials were excluded if there were no HAM-D-17 ratings available, no diagnosis of MDD, or if the study was not blinded. Trials were also excluded if HAM-D-17 assessments were not available at week 2, week 4, and at least once beyond week 4. Early improvement predicted stable response and stable remission with high sensitivity (>or= 81% and >or= 87%, respectively). Studies utilizing rapid titration vs. slow titration of mirtazapine demonstrated improved sensitivity for stable responders (98%, [95% CI = 93% to 100%] vs. 91% [95% CI = 89% to 93%]) and stable remitters (100%, [95% CI = 92% to 100%] vs. 93% [95% CI = 91% to 95%]). Negative predictive values for stable responders and stable remitters were much higher (range = 82%-100%) than positive predictive values (range = 19%-60%). These results indicate that early improvement with antidepressant medication can predict subsequent treatment outcome with high sensitivity in patients with major depressive disorder. The high negative predictive values indicate little chance of stable response or stable remission in the absence of improvement within 2 weeks. A lack of improvement during the first 2 weeks of therapy may indicate that changes in depression management should be considered earlier than conventionally thought. ©Copyright 2009 Physicians Postgraduate Press, Inc.