We examine how retention motives affect prosecutor behavior under different evaluation criteria. In particular, we analyze how prosecutors of differing capabilities respond in choosing which cases to take to trial and which to plea bargain. We show how different criteria distort the mix of cases chosen for trial and that the direction of the distortion depends crucially on the evaluation tool used. Optimal evaluation metrics are derived that combine multiple signals of performance and are shown to achieve the first‐best outcome. This article is protected by copyright. All rights reserved.