A core prediction of recent "dual-self" models is that a person's risk attitudes depend on her current level of self-control. While these models have received a lot of attention, empirical studies tailored to testing their core prediction are lacking. Using two prominent models, we derive precise hypotheses for choices between risky monetary payoffs in a state of low self-control, compared to regular self-control; in particular, lower levels of self-control should induce stronger risk aversion for stakes within a particular range. We test the hypotheses in a lab experiment with a large number of subjects (N = 308), using a well-established self-control depletion task and measuring risk attitudes via finely graduated choice lists. While independent manipulation checks document the effectiveness of our depletion task, we do not find any evidence for increased risk aversion after self-control depletion. Our findings have important implications for the future modeling of decision making under risk.