Significant variation in performance in motor imagery tasks impedes their wide adoption for brain-computer interface applications. Previous researchers have found that resting-state alpha-band power is positively correlated with MI-BCI performance. In this study, we designed a neurofeedback training protocol based on the up-regulation of the alpha band relative power to investigate its effect on MI-BCI performance. The principal finding of this study is that alpha NFT could successfully help subjects increase alpha-rhythm power and improve their MI-BCI performance. An individual difference (...) was also found in this study in that subjects who increased alpha power more had a better performance improvement. Additionally, the functional connectivity of the frontal-parietal network was found to be enhanced after alpha NFT. However, the enhancement failed to reach a significant level after multiple comparisons correction. These findings contribute to a better understanding of the neurophysiological mechanism of cognitive control through alpha regulation. (shrink)