This thesis is based on the intention to implement a motor imagery based Brain Computer Interface training paradigm for motor rehabilitation on stroke patients. The study wanted to investigate whether or not the presentation of contingent motor feedback in a motor imagery based BCI training is necessary for learning to control and modulate the SMR band power compared to a motor imagery condition with mock feedback. Fourteen healthy subjects performed four randomly presented motor related tasks in four consecutive sessions of BCI training: active and passive movements with a hand orthosis, as well as motor imagery followed or not with motor feedback via the same orthosis. Two groups were created, one receiving contingent (CF), and the other one receiving fake motor feedback (FF) in the imagery condition with included feedback. The corresponding hypothesis assumed that the group receiving CF would have stronger SMR power modulation and a better BCI control of the hand orthosis than the group receiving FF.