In recent years, a discourse is emerging in education that emphasises the study of the impact of in-service teacher education on student outcomes (more often than not student outcomes are seen in the form of test scores of academic achievement). Implicit in this discourse is the view that the impact of in-service teacher education is directly observable on students' outcomes, suggesting that the variables in a school or classroom are connected in some kind of a causal link. However, it is problematic to view variables in a social setting such as a school or a classroom as being in causal relationships because social settings are complex so that it is not possible to control the variables or the outcomes. Hence, one cannot convincingly study the outcomes without also studying the process and its complexity. In this article, I describe an action research study undertaken to study the impact in the classroom, of new teaching strategies introduced as part of an in-service teacher education programme in Karachi, Pakistan. By describing this study I mean to suggest that action research is an appropriate methodology to study the impact of in-service teacher education.