Abstract This study tests the adequacy of the axioms underlying Luce and Weber's (1986) conjoint expected risk model. Risk judgments are found to be transitive. Monotonicity or the substitution principle per se seems to hold, but the related probability accounting assumption is violated. The conjoint structure assumptions about the effect of change of scale transformations on risk hold for negative-outcome lotteries but encounter some difficulty for positive-outcome lotteries. Possible explanations for violations are suggested, and implications of these results for the modeling of perceived risk are discussed.