This study presents robust evidence on the relationship between teacher pay and turnover using detailed panel data from Texas. While controlling for changes in district and local labor market characteristics, the author estimates an overall turnover elasticity of -1.4 and shows that the effect is largest for inexperienced teachers, declines with experience, and disappears around 19 years of experience. Combining these results with what is known about the relationship between teacher value-added and experience, the author shows that paying teachers more improves student achievement through higher retention rates. The results also suggest that adopting a flat salary schedule may be a cheap way to improve student performance. The author finds no evidence that pay effects vary by the teacher's gender or subject taught.