The article reviews the research on teacher effectiveness and develops the case for a model of teacher effectiveness in which differential effectiveness is incorporated. Five problems with current concepts of teacher effectiveness are identified: undue influence of available techniques upon the concept; emphasis on school, to the detriment of teacher, effectiveness; tenuous relationship to teacher improvement; narrowness of operational definitions in research; and the development of generic, rather than differentiated, models. In addition the failure of existing models to explain variance in pupil outcome at the classroom level, the neglect of teacher self-evaluation, and the restricted measures of pupil outcomes are noted. A differential model is proposed incorporating five dimensions of difference. These refer to teacher activity, outside as well as inside the classroom; curriculum subject; pupil background factors; pupil personal characteristics; cultural and organisational contexts of teaching. The developmental functions of such a model for research and for teacher appraisal are explored. Four problems for implementing a differentiated model are raised: complexity, stakeholder expectations, values, and policy acceptability. These are considered in the light of the controversial Hay McBer model in England and of models developed in Europe and the USA in the early decades of the last century.