In two-choice tasks, the compatible mapping of left stimulus to left response and right stimulus to right response typically yields better performance than does the incompatible mapping. Nonetheless, when compatible and incompatible mappings are mixed within a block of trials, the spatial compatibility effect is eliminated. Two experiments evaluated whether the elimination of compatibility effects by mixing compatible and incompatible mappings is a general or specific phenomenon. Left-right physical locations, arrow directions, and location words were mapped to keypress responses in Experiment 1 and vocal responses in Experiment 2. With keypresses, mixing compatible and incompatible mappings eliminated the compatibility effect for physical locations and arrow directions, but enhanced it for words. With vocal responses, mixing significantly reduced the compatibility effect only for words. Overall, the mixing effects suggest that elimination or reduction of compatibility effects occurs primarily when the stimulus-response sets have both conceptual and perceptual similarity. This elimination may be due to suppression of a direct response-selection route, but to account for the full pattern of mixing effects it is also necessary to consider changes in an indirect response-selection route and the temporal activation properties of different stimulus-response sets.