In the present research, I introduce a new type of means under goal systems theory (Kruglanski et al., 2002): a costly means, which is instrumental to a focal goal but detrimental to alternative goal(s). An attributional inference similar to the augmentation effect (Kelley, 1971) may occur for costly means, suggesting that because they are detrimental to alternative goal(s), they must be especially instrumental to a focal goal. Moreover, individuals under high (vs. low) commitment to this focal goal may perceive a costly means as less extreme. Findings from Study 1 provide evidence for both hypotheses, and Study 2 showed that alternative goal primes lead to perceptions of costly means as more extreme. These findings recommend the integration of rational choice and goal systems theories, provide evidence for the augmentation effect as a heuristic tool, and highlight potential interventions to combat extremist cognition and behavior.