Pollination is a key ecological function of most terrestrial ecosystems. Decades of research on single-trophic-level communities, particularly plant communities, have helped to build the foundation of diversity–function theory. Yet as it stands, this theory appears to be less useful for intertrophic-level functions such as pollination, as evidenced by empirical findings that are often inconsistent with theoretical expectations. In this review, we evaluate how canonical diversity–function theory has been applied to pollination function, focusing on empirical studies of the mechanisms that drive pollinator diversity–function relationships. We first identified key features of the pollination function that have hampered reconciliation with the current theory. We then examined terminology for mechanisms used to discuss the findings from pollinator diversity–function studies that are sometimes inconsistent with established ecological concepts. We propose a revised diversity–function framework and describe two non-canonical diversity–function mechanisms that are particularly applicable to pollination. The first, “interactive functional complementarity,” was identified previously but remains overlooked. The second, a new diversity–function mechanism, “functional enhancement,” occurs when pollinator diversity increases within-niche activity. Finally, we discuss experimental approaches necessary to detect diversity–function effects in pollination.