Action outcomes that contingently follow our actions become bi-directionally associated with them. Thus, by anticipating desired outcomes, we are both able to select and plan corresponding actions and to proactively shift our attention towards the location of our action’s future outcome (anticipatory saccades). Such anticipatory saccades reflect a proactive effect monitoring process that prepares a later comparison of expected and actual outcome. Here, I addressed the question how we control and adjust such proactive shifts of attention (i.e., proactive effect monitoring) following unexpected outcomes. Participants’ left and right key presses caused videos of falling items to appear at predictable locations spatially compatible or incompatible with the action. In 80% of the trials, the videos appeared at the expected location and showed regular items. In 10% of the trials each, however, the videos either appeared at the unexpected location or they contained unexpected, physically-impossible content (e.g., a spoon melting upon impact). Manual response measures in trials following unexpected outcomes were only affected by preceding unexpected outcome locations. Anticipatory saccades, however, showed dissociable effects of unexpected location as compared to unexpected content. Following a trial with an unexpected outcome location, participants were less likely to anticipatorily saccade towards the frequent outcome location as compared to following an expected outcome. Conversely, they were more likely to anticipatorily saccade towards the future outcome after encountering unexpected outcome content rather than an expected outcome. This suggests that cognitive control processes selectively adapt anticipatory saccades (i.e., proactive effect monitoring) depending on the nature of unexpected events.