Motor imagery (MI) relies on conscious mental simulation of a motor act without overt motor output and can promote motor skill acquisition and facilitate rehabilitation for patients with stroke or neurological conditions. Although a plethora of neuroimaging studies have investigated the neural network of MI regarding different body parts, exploration of the neural correlates to facial MI remains warranted. Here, we used functional MRI with a large cohort of 41 participants who underwent motor execution (ME) and MI runs of mouth-stretching tasks. Then, we carried out conjunction and contrast analyses to investigate the commonalities and differences between the two conditions. Conjunction analysis, representing the shared neural network between ME and MI, showed activation in the primary motor cortex, primary and secondary somatosensory cortices, premotor cortex, parietal lobe, anterior insula, supplementary motor area (SMA) and pre-SMA, thalamus, putamen, and caudate. Contrast analysis showed greater activation of primary motor cortex, primary and secondary somatosensory cortices, SMA, anterior insula, and the thalamus in response to ME than MI and greater activation of the premotor cortex, pre-SMA, putamen, and caudate in response to MI than ME. Interestingly, we found exclusive activation in the anterior cingulate cortex and left ventrolateral prefrontal cortex in response to MI, reflecting the motor inhibition network responsible for blocking the transmission of motor commands to peripheral effectors during mental rehearsal. Taken together, these findings show that, despite the neural overlap between ME and MI, there are different degrees of activation within this overlap, and that MI normally involves motor command inhibition possibly mediated by the anterior cingulate cortex and ventrolateral prefrontal cortex.