The effects of a pharmacological preventive treatment can be described by two conditional rules: the rule "about efficacy" and the rule "about noxiousness". On the basis of a difference in the logical structure of the two rules we hypothesize that individuals seeking support for the treatment will focus more on falsificatory evidence concerning efficacy and on confirmatory evidence concerning noxiousness. The results of 3 experiments using the Wason selection task (Experiments 1 and 3) and an evaluation of contingency tables task (experiment 2) confirm the hypothesis. The effect is not caused by the linguistic structure of the two rules, but springs from a mechanism of focusing by relevance that we interpret in the light of the mental model theory. The effect can contribute to the long-term loss of confidence toward preventive treatments, thus prompting non-adherence behaviors. This is particularly true for preventive treatments with many side effects, as in the immunosuppression following a transplant.