That summative assessment drives learning has become one of the most frequently stated maxims in the literature of assessment and learning in higher education. However, a careful review of the empirical research often cited in support of this proposition may cause us to reconsider its veracity and to seek a more nuanced understanding of the research findings. Seminal research on the influence of assessment on learning in higher education was reviewed on the basis of its context, research methods used, sampling, reported findings, and the generalisability of these findings. This study found that the treatment of the research reported in frequently cited works such as Making the grade, The hidden curriculum, and Up to the mark has often oversimplified, and thus misrepresented, the research findings, leading to singular interpretations of complex, multi-faceted phenomena. Other research suggesting limitations to the capacity of assessment per se to improve students’ approaches to learning is often misunderstood or under-emphasised, leading to the risk of exaggerated claims for the capacity of ‘alternative’ forms of assessment to foster effective learning processes in students. The findings of this review lead to a proposed agenda for empirical research to address what seem to be serious gaps in our understanding of fundamental aspects of the interactions between assessment and learning.