We study the removal of information that favors one group over another. Empirically, we focus on banning the use of credit reports to screen job applicants, a practice alleged to disadvantage minority groups with poorer average credit. We estimate these bans decrease job-finding rates for blacks by 3 percentage points and increase involuntary separations for black new hires by 4 percentage points. Theoretically, we show the incidence of removing information depends on the information’s relative precision, not necessarily group averages. We estimate credit checks benefit black job seekers because other screening tools, such as interviews, are noisier signals for blacks.