Involvement has long been theoretically specified as a crucial factor determining the persuasive impact of messages. In social judgment theory, ego-involvement makes people more resistant to persuasion, whereas in dual-process models, high-involvement people are susceptible to persuasion when argument quality is high. It is argued that these disparate predictions might be reconciled by either different involvement types (i.e., value relevant vs. outcome relevant) or different attitude modification processes (i.e., attitude change vs. attitude formation). An experiment (N = 684) varying topic, position advocated, outcome relevance, and argument quality tested these moderators. The data were consistent with existence of two different types of involvement, but none of the theoretical predictions were consistent with the data. Instead, a main effect for argument quality had the largest impact on attitude change. Regardless of value-relevant involvement, outcome-relevant involvement, and attitude modification process, participants were more persuaded by high- rather than low-quality arguments, with boomerang effects observed for low-quality arguments. These findings highlight the importance of sound message design in persuasion. © 2007 International Communication Association.