Abstract: This paper provides a theoretical and empirical analysis of how judicial election systems affect performance via both selection of different types of judges and by imposing incentive effects of retention elections. We use panel data on state supreme courts between 1947 and 1994 and exploit variation in the procedures for selection and retention of judges over time, as well as the staggered judicial electoral cycle. When judges are up for election in competitive systems, performance falls -- but not in systems with weak electoral incentives. Our results are consistent with the view that competitive electoral systems tend to reduce judge performance, and that nonpartisan judicial elections are more competitive than partisan elections. Moving from a competitive election system to a less competitive election system increases judge performance, and vice versa. Moving from electoral selection by voters to merit selection by an expert commission results in the selection of higher-performing judges.