The combination of a lower-limb exoskeleton with brain computer interfaces (BCI) can assist patients with motor impairment to walk again. In addition, it can promote the neural plasticity of the affected brain region. The present paper shows a research performed on seven able-bodied subjects that walked with an assistive exoskeleton controlled by external commands. The main objective was to identify in which frequency band the differences between periods of motor imagery and rest were more evident. The comparison was done with different classifiers and the results reveal that for the majority of them, the frequency band of 14–19 Hz provided the highest accuracy.