In this speech the author considers the problem of measuring teacher effectiveness from the point of view of an educational researcher. He examines two prevalent methodological approaches: evaluating the process and evaluating the product. The first approach involves the description of a study which concluded that teachers rated high by supervisors and well-liked by pupils were not the teachers whose pupils showed the greatest gains and who judged themselves most effective. The second approach reviews the problems of the reliability and validity of tests and advocates giving up the idea of measuring teacher effectiveness in terms of pupil gains on tests. The recent work of Flanders, Bellack, Smith, Hughes and others is commented on, with the conclusion that the more dimensions of teacher behavior that can be measured and correlated with pupil gains, the more likely is the discovery of one or more correlating with teacher effectiveness. Reviewing common practices that measure teacher effectiveness, the author points out that educational research does not support Dewey's "we learn by doing," or Skinner's reinforcement theories. Instead, less obvious things, such as the cognitive level of questions asked, probing questions, clarity and structure, should be looked for in measuring teacher effectiveness. (JA)