Research on teaching does not play a significant role in the practice of teaching in schools despite the fact that the first scientific studies of teaching were reported nearly 90 years ago. Although there is clear evidence that practical classroom experience unaided by research is not a sufficient basis for effective teaching, it is also true that research on teaching has been unable to provide a more adequate basis for effective teaching. A brief history of research on teaching shows that, after having gone down several "blind alleys," such research has emerged with a clear focus on direct data gathering in the classroom itself. The last 30 years have seen considerable controversy about the appropriate methods for research on teaching, but what is needed to resolve questions of method and purpose is a clear picture of the kind of knowledge that research on teaching should produce. The paper describes a model of the kind of pragmatic knowledge that research should be producing if it is to be genuinely useful to teachers. The model includes both the theoretical and the practical dimensions of teaching in a single structure, tactics and monitoring signs that teachers need to run an effective program, and an account of the underlying learning and social processes connecting these tactics to the outcomes of classroom activities. (JD)