Action-feedback monitoring is essential to ensure meaningful interactions with the external world. This process involves generating efference copy-based sensory predictions and comparing these with the actual action-feedback. As neural correlates of comparator processes, previous fMRI studies have provided heterogeneous results, including the cerebellum, angular and middle temporal gyrus. However, these studies usually comprised only self-generated actions. Therefore, they might have induced not only action-based prediction errors, but also general sensory mismatch errors. Here, we aimed to disentangle these processes using a custom-made fMRI-compatible movement device, generating active and passive hand movements with identical sensory feedback. Online visual feedback of the hand was presented with a variable delay. Participants had to judge whether the feedback was delayed. Activity in the right cerebellum correlated more positively with delay in active than in passive trials. Interestingly, we also observed activation in the angular and middle temporal gyri, but across both active and passive conditions. This suggests that the cerebellum is a comparator area specific to voluntary action, whereas angular and middle temporal gyri seem to detect more general intersensory conflict. Correlations with behavior and cerebellar activity nevertheless suggest involvement of these temporoparietal areas in processing and awareness of temporal discrepancies in action-feedback monitoring.