Mutualistic interactions among free-living species have shaped much of biodiversity on Earth. Ironically, however, mutualism has not been prominently featured in theoretical ecology. Recent efforts have tried to fill this gap by assessing to what degree the structure of plant-animal mutualistic networks affects species persistence. Here, we review this growing literature emphasizing how different papers relate to each other and to what extent their conclusions depend on particular assumptions. A central concept in this effort is that of structural stability. While main approaches in theoretical ecology focus on local dynamical stability, structural stability shifts the question to how large is the range of parameter values compatible with the stable coexistence of all species. This structural approach has shown that mutualism has to be understood as a balance to competition and that network architecture should be seen as affected by both stability and feasibility constraints. Constraints on the dynamical stability of these communities set up a maximum mutualistic strength. Constraints on their feasibility tend to push interaction strength near this limit and select for a nested architecture. These results, however, assume similar interspecific competition amongst species and very small mutualistic strength. Future work should thus explore the consequences of relaxing these constraints.