Attention is multifaceted, with evidence for distinct mechanisms of attentional facilitation and suppression processes. Interestingly, much less is known about the spatial coordinate system of suppression compared to that of facilitation. The present study examined the coordinate system of spatial suppression by manipulating gaze position and distractor regularities, asking whether suppression is coded in retinotopic (eye-centered) and/or spatiotopic (world-centered) coordinates, and if this varies with more ecological and dynamic contexts. In the current study, we demonstrate that learned spatial suppression primarily transfers across gaze position in retinotopic coordinates; however, in more dynamic contexts favoring spatiotopic information, spatial suppression can be learned in spatiotopic coordinates. These results suggest that the default coordinate system of spatial suppression is retinotopic, but suppression can be rapidly learned in spatiotopic coordinates when a spatiotopic representation is beneficial in dynamic contexts.