This article reviews evidence regarding the intertemporal reliability of teacher rankings based on value-added methods. Value-added methods exhibit low reliability, yet are broadly supported by prominent educational researchers and are increasingly being used to evaluate and fire teachers. The article then presents a cost-effectiveness analysis suggesting that the use of value-added methods to identify and fire the bottom 40% of all teachers reduces average student achievement, is extremely expensive, and is not cost-effective. The policy implication of these results is discussed.