A Brain-Computer Interface (BCI) represents the ultimate means of communication for people with severe paralyses or who are in a locked-in state. However, the usage of BCI is still severely limited in terms of accuracy and performance speed. One possible way to overcome these restrictions would be the detection of errors after incorrect events in the electroencephalogram (EEG). In this study 13 subjects participated in a first experiment to provide data for offline analysis of interaction error potentials (ErrPs) which were recorded after observation of falsely interpreted user-commands by an interface. These characteristic waveforms were later used to classify errors in online experiments combined with motor imagery (MI). Here, the detection of false movements could improve the accuracy significantly.