Do judges selected by merit review commissions perform better than elected judges or those directly appointed by elected officials? This is a central question in both the academic study of state judicial institutions and the policy discourse about how to reform them. To address it, we take advantage of the variation in the means of the selection for trial court judges within Arizona, a state comprised of appointed, elected, and merit-selected trial court judges. This unique context allows us to use an objective measure of judicial performance – the reversal rate of trial court cases appealed to Arizona’s state appellate courts – to evaluate judges by their means of selection. We gather an original dataset on 2919 cases heard by 176 judges, estimating multivariate models that control for characteristics of cases and of judges. Overall, we find that elected judges have a lower reversal rate than merit-selected judges. Our findings question the conventional wisdom in the state courts literature in favor of merit selection and against judicial elections, and encourage further work on the effects of judges’ means of selection beyond state supreme courts to include state appellate and trial courts.