Well-documented racial disparities in rates of exclusionary discipline may arise from differences in unobservable student behavior or from bias, in which treatment for the same behavior varies by student race or ethnicity. We provide evidence for the presence of bias in school discipline decisions using statewide administrative data that contain rich details on individual disciplinary infractions. Two complementary empirical strategies find racial differences suggestive of bias in suspension outcomes. The first uses within-incident variation in disciplinary outcomes across White, Black, and Hispanic students. The second employs individual fixed effects to examine how consequences vary for students across incidents based on the race of the other student involved in the incident. Both approaches find that Black students face higher suspension probabilities and longer suspensions than White students and are suspended for longer than Hispanic students. There is no evidence of Hispanic-White disparities. The similarity of findings across approaches and the ability of individual fixed effect models to account for unobserved characteristics common across disciplinary incidents provide support that remaining racial disparities are unlikely to be driven by differences in behavior.