In two experiments, participants were asked to rate the riskiness of three imaginary diseases after reading a short description of each, varying in controllability and severity, two of the most important determinants of risk perception. Cognitive load was also manipulated in Experiment 1, to check whether any of the two factors can be moderated by System 2 processes. The main effects of controllability and severity and the interaction effects of the two were significant. The simple effect of severity was larger in the less severe conditions. However, the main effect or the interaction effects of cognitive load was not significant. In Experiment 2, participants were primed to do system 2 processing by direct instruction or indirect priming. None of the two priming had significant effects. Results of the two experiments were interpreted to suggest that controllability and severity affect risk perception in an automatic manner.