The present study was concerned with the effects of education level and area of expertise on performance on the standard abstract selection task. Subjects had received bachelor’s or doctoral degrees. Contrary to some recent results reported by Hoch and Tschirgi (1985), no effect of education level was found. However, there was a significant effect for area of expertise. Subjects trained in mathematics performed better than subjects from computer science, electrical engineering, and the social sciences. An explanation in terms of the mathematics subjects’ greater likelihood of using a disconfirmation strategy and greater familiarity with the relevant propositional logic is offered.