Previous research has shown that when confronted with failure, individuals with a fixed view of intelligence (entity theorists) perform worse on subsequent tasks than those with a malleable view of intelligence (incremental theorists). This study finds that entity theorists perform worse than incremental theorists only when they believe that a subsequent task measures the same ability as the task they previously failed. However, when individuals believe that the subsequent task measures an ability unrelated to the ability needed for the initial failed task, incremental theorists perform worse than entity theorists. Across five studies, we show that entity theorists are more likely to choose a different-ability task as a second task and perform better than incremental theorists on that task. We also examine the role of thoughts about previous failure in the performance differences.