We report two studies investigating how naive reasoners evaluate the probability that a conditional assertion is true, and the conditional probability that the consequent of the conditional is true given that the antecedent is true. The mental model theory predicts that individuals should evaluate the probability of a conditional on the basis of the mental models representing the conditional, and that evaluations calling for a greater number of models should be more difficult. It follows that the probability of a conditional should differ from the corresponding conditional probability. The results of the studies corroborated these predictions, and contrast with alternative accounts of naive evaluations of the probability of conditionals.