Task conflict has been the subject of a long-standing debate in the literature—when does task conflict help or hurt team performance? We propose that this debate can be resolved by taking a more precise view of how task conflicts are perceived in teams. Specifically, we propose that in teams, when a few team members perceive a high level of task disagreement while a majority of others perceive low levels of task disagreement—that is, there is positively skewed task conflict, task conflict is most likely to live up to its purported benefits for team performance. In our first study of student teams engaged in a business decision game, we find support for the positive relationship between skewed task conflict and team performance. In our second field study of teams in a financial corporation, we find that the relationship between positively skewed task conflict and supervisor ratings of team performance is mediated by reflective communication within the team.