Should individuals delegate thinking about complex choice problems to the unconscious? We tested two boundary conditions on this suggestion. First, we found that in a decision environment similar to those studied previously, self-paced conscious thought and unconscious thought had similar advantages over conscious thought constrained to a long fixed time interval in terms of identifying the option with the highest number of positive outcomes. Second, we found that self-paced conscious thought performed better than unconscious thought in a second decision environment where performance depended to a greater extent on magnitudes of the attributes. Thus, we argue that it is critical to take into account the interaction of forms of processing with task demands (choice environments) when considering how to approach complex choice problems.