Relatively little research has been conducted into designing interfaces that allow GIS users to interact effectively with geospatial data in mobile environments. Users on the move are faced with limited interaction modalities. The standard mode of input on mobile devices is the pen or stylus, which some users may find difficult or too time-consuming to use. Voice commands, combined with pen input, can provide an attractive alternative for interacting with mobile GIS, as speech is a natural form of interaction. However, the idea of combining speech and pen input in mobile GIS is relatively unexplored. To this effect, we have developed a multimodal interface to a mobile GIS, providing users with the ability to freely switch between modalities to suit current tasks or environments.