A new capacity to track the inputs and outcomes of individual students' education production function has spurred a growing number of school districts to attempt to measure the productivity of their teachers in terms of student outcomes. The use of these value-added measures of teacher effectiveness is at the center of current education reform. This study links the technical work of academic researchers with the implementation and policy considerations school districts are likely to face in incorporating value-added measures in their teacher evaluations. First, I assess the choices the district must make in specifying one or more models. Then, I evaluate three potential threats to the validity of the inferences from value-added data: the influence of prior inputs in a student's education production function, ceiling effects in the test instrument, and the sorting of students to teachers. I end by considering to what extent value-added measures could be useful to districts in monitoring the distribution of effective teachers to its students and personnel decisions such as retention and compensation.