We investigate the consistency and stability of individual risk preferences by manipulating cognitive resources. Participants are randomly assigned to an experiment session at a preferred time of day relative to their diurnal preference (circadian matched) or at a non-preferred time (circadian mismatched) and choose allocations between two risky assets (using the Choi et al., 2007, design). Consistency of behavior of circadian matched and mismatched subjects is statistically the same, however mismatched subjects tend to take more risks. We conclude that, consistent with several theories, preferences are rational yet can change depending on state-level cognitive resources.