As human‐aided range expansions and climate change alter the distributions of plants and their herbivores, predicting and addressing novel species interactions will become increasingly pressing for community ecologists. In this context, a key, surprisingly understudied question is: when an exotic plant is introduced, which herbivores will adopt this new potential host? Whether the plant is a weed, an ornamental, or a crop, the development versus non‐development of a novel plant–insect interaction can have profound effects for both economic and conservation applications. In this paper, we sketch mechanistic and statistical frameworks for predicting these interactions, based on how plant and herbivore traits as well as shared evolutionary history can influence detection, recognition, and digestion of novel plants. By emphasizing mechanisms at each of these steps, we hope to clarify different aspects of novel interactions and why they may or may not occur. We also emphasize prediction and forecasting, as a major goal is to know in advance which interactions will develop from the many plant or insect introductions that occur in natural and man‐made systems.