Mobile multimodal systems raise some novel usability challenges. Some of these are due to the combination of two characteristics of mobile systems and multimodal systems, respectively: the competition between the system and the environment for the user’s attention; and the availability of multiple modalities for user input and system output. This paper first presents a theoretical argument that the set of problems raised by the combination of these two characteristics is more than just the union of the two sets raised by each characteristic itself. It then discusses one relatively new method - mobile eye tracking - that can help with the empirical study of these problems. Finally, it considers the question of how automatic system adaptation to a user’s current resource limitations might ultimately enhance the usability of mobile multimodal systems.