Accumulating evidence suggests that the overall level of cortical activation controlling a voluntary motor task that leads to significant muscle fatigue does not decrease as much as the activation level of the motoneuron pool projecting to the muscle. One possible explanation for this "muscle fatigue>cortical fatigue" phenomenon is that the brain is an organ with built-in redundancies: it has multiple motor centers and parallel pathways, and the center of activation may shift from one location to another when neurons in the previous location become fatigued. This hypothesis was tested by estimating the changes of source locations of high-density (64 channels) scalp electroencephalographic (EEG) signals collected during both fatigue and non-fatigue motor tasks. A current dipole model was used to estimate the EEG sources. The fatigue motor task induced significant muscle fatigue, and the non-fatigue task did not. The EEG signal source that indicated the center of brain activation showed substantial location shifts during the fatigue motor task. The shifts could not be explained by variations of source locations caused by error estimated from the non-fatigue task EEG and simulated data. Compared to the non-fatigue condition, the weightedcenter of the source locations for all the participants shifted toward the right hemisphere (ipsilateral to the muscle activation), anterior, and inferior cortical regions under the fatigue condition. Fatigue did not alter dipole (source-signal) strength or the overall level of brain activation. The brain may avoid fatigue by shifting neuron populations that participate in a fatiguing motor task.