In many interspecific interactions, the balance of costs and benefits varies with ecological circumstances. As a prominent example, seed-caching granivores may act as seed predators and reduce plant recruitment or as seed dispersers and increase recruitment, making it difficult to interpret whether differences in seed removal by granivores would harm or benefit plant populations. We used a heuristic model to evaluate the outcome of plant-granivore interactions, using commonly measured field data: probability of seedling emergence when granivores are excluded, and emergence of cached and uneaten seeds. Published studies to date suggest that the outcome of plant-rodent interactions tends weakly towards mutualism, but differs among particular plant–granivore pairs and ecological conditions, supporting the notion of context-dependence. A modeling framework also allowed us to distinguish parameters that affect the qualitative outcome of plant–granivore interactions from those that do not. Similar approaches would facilitate more efficient and cost-effective evaluation of complex species interactions.