Ontogenetic variation is the most fundamental biological aspect of an organism, and stage-structured prey–predator relationships have increasingly been studied for better understanding food webs. However, little is known about stage-structured mutualism and its community consequences. Here, I aim to argue that mutualism can be viewed as inherently stage-structured while drawing research attention to the little-studied issue. A synthesis of the literature across different mutualisms (e.g. pollination, seed dispersal, nutritional, and defensive) demonstrates that mutualism is commonly stage-structured. In addition, I present a general theoretical framework for describing community dynamics mediated by stage-structured mutualism. It can provide testable hypotheses regarding how stage-specific partners can affect each other through the life history of a shared host and how they jointly determine the lifetime fitness of the host. For the establishment of the ontogenetic perspective of mutualism, both empirical and theoretical efforts are needed to collect and incorporate individual-level interaction data into community ecology theories, which will provide valuable insights not only into mutualism-mediated community dynamics but also into biodiversity conservation and ecosystem management.