Repetitive transcranial magnetic stimulation (rTMS) is a brain stimulation technique which had recently been investigated as a putative antidepressant intervention. However, there is little agreement about clinically useful predictors of rTMS outcome. Therefore, the objective of the present study was to determine whether specific biographical, clinical, and psychopathological parameters are associated with the antidepressant response to rTMS in a large sample of 70 depressive patients. We performed a logistic regression analysis in 70 patients with major depressive disorder treated with rTMS of the left dorsolateral prefrontal cortex testing the predictive value of various domains of the depression syndrome as well as the variables episode duration, degree of treatment resistance, and CORE criteria. Response was defined as a 50% reduction of the initial Hamilton score (HAMD). After two weeks of treatment, 21% of the patients showed a response to rTMS. The binary logistic regression model correctly assigned 86.7% of the responders and 96.4% of the non-responders to their final response group. In the model, a high level of sleep disturbances was a significant predictor for treatment response to rTMS. Also, a low score of treatment resistance and a short duration of episode were positive predictors. These findings provide new evidence that especially pronounced sleep disturbances may be a significant clinical predictor of a response to rTMS. Prospective rTMS studies are necessary to validate the predictive value of the derived model.