An experiment was performed to determine the relationship between the importance of attitudes and the extent to which they are changed by the forced compliance procedure. Subjects freely consented to write a counterattitudinal essay concerning one of several political issues. The importance of the issue assigned to subjects was systematically varied across six conditions. As predicted by McGuire's information-processing theory (1968a, 1968b), importance affected attitude change in a quadratic manner, the greatest change occurring in the conditions of moderate importance. Further analysis suggests that this inverted-U effect was not due to variations in initial attitude extremity or essay quality. Linear and nonlinear relationships were discussed.