With global attention currently focused on the challenge of providing Education for All (UNESCO 2000), we must ensure that the language of teaching and learning remains a topic on the agenda towards making sure that the education being provided is effective. This is therefore a critical time to review medium of instruction debates, and to reassess what empirical evidence exists to guide policymaking that is both appropriate and achievable. Contributing to this endeavour, this paper argues for the approximate replication of two key studies. The first is Afolayan (1976), a widely-cited study conducted in Ife, Nigeria to test the effectiveness of teaching children in the mother tongue for six years of primary education. We argue that the frequency with which the success of this study is cited, without due attention paid to the details of its methodological procedure, may actually be detrimental to the success of other experiments, thus necessitating the careful replication of the original study. The second study is Siegel (1997b), one of the few studies that have been conducted to evaluate the impact of initial education in an English-based pidgin on the subsequent learning of English. We argue that there is an urgent need for replication of one of the few available studies of pidgins and creoles in education, given the prevalence of negative attitudes towards this category of languages.