Studies of embodied cognition have demonstrated the engagement of the motor system when people process action-related words and concepts. However, research using transcranial magnetic stimulation (TMS) to examine linguistic modulation in primary motor cortex has produced inconsistent results. Some studies report that action words produce an increase in corticospinal excitability; others, a decrease. Given the differences in methodology and modality, we re-examined this issue, comparing conditions in which participants either read or listened to the same set of action words. In separate blocks of trials, participants were presented with lists of words in the visual and auditory modality, and a TMS pulse was applied over left motor cortex, either 150 or 300 ms after the word onset. Motor evoked potentials (MEPs) elicited were larger following the presentation of action words compared with control words. However, this effect was only observed when the words were presented visually; no changes in MEPs were found when the words were presented auditorily. A review of the TMS literature on action word processing reveals a similar modality effect on corticospinal excitability. We discuss different hypotheses that might account for this differential modulation of action semantics by vision and audition.