This paper examines the effects of a comprehensive performance pay program implemented in high-need schools using administrative data from Louisiana. Exploiting the within-student variation in the timing of implementation, we find a large and significant effect beginning with the second year of the program on math achievement. Similar but generally insignificant point estimates are observed in social studies. As for English Language Arts and science, there are no effects of the program. We provide evidence against changes in the composition of teacher workforce and postadoption student sorting as potential explanations for our results. We then show aggregate findings from a web-survey that teachers may have altered their teaching practices following performance-based compensation. Finally, under certain assumptions, we calculate the net present value of earnings impact to be close to a million dollars per school-year for math.