Judicial performance evaluations (JPEs) are an important part of the judicial selection process in the states, particularly those using a version of the merit plan. All states that use JPEs follow the ABA’s Guidelines (1985), which claim to minimize the potential for unconscious bias through the use of behavior-based evaluation. But these measures have yet to be subjected to rigorous analysis. This analysis of the “Judging the Judges” survey of Nevada attorneys provides such an analysis. After controlling for objective measures of judicial performance, gender and race still contribute significantly to the scores on all of the behavior-based measures implemented in the Nevada poll. I find evidence of significant unconscious bias, as social cognition theory would predict. The analysis also cast serious doubt on the overall validity of these measures of judicial quality. This result raises serious questions about the validity and fairness of JPEs around the country.