Transcranial direct current stimulation (tDCS), a noninvasive neuromodulation technology, has been found to improve motor learning. Motor-imagery based Brain-Computer Interfaces (BCIs) utilize a subject's imagined movement to control virtual and physical objects. A similar sensorimotor network active during motor movement is active during motor imagery. With this parallel between motor movement and imagery, we hypothesize that anodal tDCS can be used to improve the learning of motor imagery based BCIs. Two sets of experiments were performed: anodal/cathodal/sham stimulation using a 4 × 1 HD-tDCS combination over the left sensorimotor cortex during motor imagery performance without feedback and anodal vs. sham stimulation with feedback during BCI performance. We found a significant interaction effect on electrode C3, left sensorimotor cortex, event-related activity for stimulation type and the time block (prestim, during stim, immediately post-stim, and 30 minutes post-stim) within a single session for the motor imagery experiment. We also found a significant interaction effect of event-related activity between anode and sham stimulation during BCI learning within a single session in electrode C3, but not C4. In addition, over the course of 3 sessions of BCI learning, subjects who underwent anodal stimulation tended to have higher accuracy compared to those who had the sham stimulation (71% vs. 56% valid correct). These results illustrate that acutely, within a single session, the type of tDCS stimulation alters the neural activity underlying motor imagery and may improve motor imagery learning.