Ss, holding either extreme or moderate initial attitudes, wrote counterattitudinal essays in a test of contradictory hypotheses derived from Festinger’s cognitive dissonance theory and Bem’s self-perception theory. The results indicated, as predicted by dissonance theory, that Ss holding extreme initial attitudes showed more attitude change after counterattitudinal advocacy than Ss holding moderate initial attitudes. It was demonstrated that the results were not due to regression effects, to the production of differentially persuasive essays across the extremity conditions, or to the fact (...) that Ss holding extreme attitudes had more ’’room for. (shrink)