Many public school districts are beginning to implement performance pay programs that provide teachers the opportunity to earn pay bonuses based on measures of teaching performance. Despite the growing number of districts offering these programs, we know little about their effects. Theory suggests that performance pay programs may provide an incentive for teachers to work harder (an effort effect). In addition, districts offering performance pay may attract teachers with higher average ability (a selection effect). Empirical work investigating the existence of these effects is mixed in the case of the effort effect and nonexistent in the case of the selection effect. This study is the first to attempt to empirically test for the existence of a teacher selection effect resulting from performance pay programs. I show that the existence of a selection effect may be revealed in differences in total pay and bonus probabilities between teachers who self-select into performance pay programs and teachers who are exogenously assigned. If self-selectors earn a higher expected total pay or are more likely to earn a bonus than an exogenously assigned teacher, I show that this implies that self-selectors are higher ability teachers on average. I test for this difference in cross-sectional data from the Schools and Staffing Survey (SASS) and in longitudinal data from a single performance pay district in Minnesota. In each case, I fail to find evidence of a selection effect. In the cross-sectional analysis, I find that while self-selectors earn a higher expected total pay, they are also less likely to earn a performance bonus. In the longitudinal analysis, I find that teachers who joined a district in the few years prior to its adoption of performance pay are not measurably different from teachers who joined in the years after adoption. Post performance pay joiners are not measurably different in terms of their education, experience, or likelihood of earning a performance bonus. While I fail to find evidence of a selection effect, that should not be taken as proof that performance pay programs, in general, do not produce their advertised benefits. The SASS analysis relies on several strong assumptions that potentially undermine the credibility of the results. For the analysis of the performance pay district in Minnesota, this study’s inability to find evidence of a selection effect is likely a result of the district’s high relative base compensation and nearly guaranteed bonus award.