Police officers have a high degree of discretion in how they conduct themselves on duty. This is caused in part by the highly localized regulation of policing and the inherent vagueness of the central legal construct governing police contact with civilians—the “reasonable suspicion” standard. Psychological science has demonstrated time and again the many ways that decision-making under uncertainty is vulnerable to errors arising from biases and heuristics. High discretion (i.e., wide latitude in how to act) broadens the set of conditions under which one can exercise prerogatives, thereby increasing uncertainty. Because of the pervasiveness of race–crime stereotypes, police, like all of us, are likely to judge and act toward minorities with greater suspicion. Experimental and administrative data bear this out. Short-term training to mitigate the resultant disparities in stops, searches, and use of force is unlikely to be effective because people need chronic motives, feasible strategies, and sufficient opportunities to override biases. Until reliable methods to overcome biases can be developed, structural reform must be considered. Policing data consistently shows that disparities are greatest under high discretion. Constraints on discretion have been shown to increase accuracy and reduce disparities. Methods to operationalize these constraints in ways to promote fairness and public safety must be developed, tested, and implemented.