The shortcomings of the three main approaches to measuring teacher effectiveness (systematic observation, participant observation, and questionnaires) are discussed and problems of research design and analysis considered. The development of better measures of educational benefits to complement the emphasis on test scores is suggested. (Author/SK)