Despite decades of study and enormous effort, we know little about how to train or select high quality teachers. We do know, however, that there are huge differences in the effectiveness of classroom teachers and that these differences can be observed. Given this situation, the path to improvement rests with enhanced evaluation systems for teachers combined with better personnel systems that link retention and reward to effectiveness. Without knowing the background, preparation or attributes that make a good teacher, we cannot rely on the credentialing process to regulate the quality of people who enter the profession. Therefore the most sensible approach is to expand the pool of potential teachers but tighten up on decisions about retention, tenure and rewards for staying in teaching. Many states have already begun to endorse alternative routes into teaching so that there is a wider pool. We will still make the best judgment we can about who will do well, but we simply have to recognize that mistakes happen. We then need to make active decisions about who to retain and who not to retain. Evaluation of teacher performance becomes key. Gains in student achievement should be one element, because improving student achievement is what we are trying to do, but this is not even possible for most teachers. Moreover, nobody believes that decisions should be made just on test scores. What we need is some combination of supervisor judgments with the input of professional evaluators. Public schools in Washington, D.C., provide evidence that improved personnel policies work. In Washington, a sophisticated and reliable evaluation system for teachers is used to make both pay and retention decisions. The best teachers get large rewards including substantially higher base salaries. The worst teachers are dismissed. And evaluations of this program indicate that it is working and that student achievement is rising. The problem is that unions and schools resist active personnel decisions – to the detriment of students.