Background: Major depressive disorder afflicts an estimated 17% of individuals during their lifetimes at tremendous suffering and costs. Psychodynamic therapy may be a treatment option for depression, but the effects have only been limitedly assessed in systematic reviews. Method: Using Cochrane systematic review methodology, we compared the benefits and harms of psychodynamic therapy versus 'no intervention' or sham for major depressive disorder. We accepted any co-intervention, including antidepressants, as long as it was delivered similarly in both intervention groups. Trials were identified by searching the Cochrane Library's CENTRAL, MEDLINE via PubMed, EMBASE, Psychlit, Psyc Info, and Science Citation Index Expanded until February 2010. Two authors independently extracted data. We evaluated risk of bias to control for systematic errors. We conducted trial sequential analysis to control for random errors. Results: We included five trials randomizing a total of 365 participants who all received antidepressants as co-intervention. All trials had high risk of bias. Four trials assessed 'interpersonal psychotherapy' and one trial 'short psychodynamic supportive psychotherapy'. Meta-analysis showed that psychodynamic therapy significantly reduced depressive symptoms on the 17-item Hamilton Rating Scale for Depression (mean difference - 3.01 (95% confidence interval - 3.98 to - 2.03; P < 0.00001), no significant heterogeneity between trials) compared with 'no intervention'. Trial sequential analysis confirmed this result. Limitations: Our results are based on few trials with high risk of bias and a limited number of participants so our results may be questionable. Conclusions: Adding psychodynamic therapy to antidepressants might benefit depressed patients, but the possible treatment effect measured on the Hamilton Rating Scale for Depression is small.