The neural bases of verbal (nonspatial) working memory (VWM) have been primarily examined using visual stimuli. Few studies have investigated the neural bases of VWM using auditory stimuli, and fewer have explored modality differences in VWM. In this study, we used functional magnetic resonance imaging (fMRI) to examine similarities and differences between visual VWM (vis-VWM) and auditory VWM (aud-VWM) utilizing identical stimuli and a within-subjects design. Performance levels were similar in the two modalities and there was extensive overlap of activation bilaterally in the dorsolateral and ventrolateral prefrontal cortex (DLPFC and VLPFC), intraparietal sulcus, supramarginal gyrus and the basal ganglia. However, a direct statistical comparison revealed significant modality differences: the left posterior parietal cortex, primarily along the intraparietal sulcus, showed greater responses during vis-VWM whereas the left dorsolateral prefrontal cortex showed greater responses during aud-VWM. No such differences were observed in the right hemisphere. Other modality differences in VWM were also observed, but they were associated with relative decreases in activation. In particular, we detected bilateral suppression of the superior and middle temporal (auditory) cortex during vis-VWM, and of the occipital (visual) cortex during aud-VWM, thus suggesting that cross-modal inhibitory processes may help to provide preferential access to high-order heteromodal association areas. Taken together, our findings suggest that although similar prefrontal and parietal regions are involved in aud-VWM and vis-VWM, there are important modality differences in the way neural signals are generated, processed and routed during VWM.