When are U.S. Supreme Court justices more likely to recuse themselves from cases? This article proposes a strategic model of recusal behavior, hypothesizing that the justices balance statutory guidelines concerning recusals against other policy and institutional goals. Using data from the Supreme Court Database, I find evidence that recusal behavior is influenced by a combination of statutory, policy, and institutional considerations. Consistent with statutory explanations, which emphasize the elimination of bias or its appearance, justices are more likely to recuse themselves from cases when business interests are before the Court, when they have served for shorter terms, and when they have previously acted as Solicitor General. However, I also find that the justices are less likely to recuse themselves when cases are likely to be close or when the justices' policy goals are likely to be advanced by participating. These findings suggest that while the justices do follow statutory recusal guidelines, they also have other institutional and policy incentives that lead them to participate in cases despite their conflicts of interest.