Dynamic interactive maps with powerful interface capabilities are beginning to emerge for a variety of geographical information systems, including ones situated on portables for travelers, students, business and service people, and others working in field settings. In part through the design of more expressive and flexible input capabilities, these map systems can provide new capabilities not supported by conventional interfaces of the past. In this research, interfaces supporting spoken, pen-based, and multimodal input were analyzed for their effectiveness in interacting with map systems. Input modality and map display format were varied as people completed realistic tasks with a simulated map system. The results identified a constellation of performance difficulties with speech-only map interactions, including elevated performance errors, lengthier task completion time, and more complex and disfluent input—problems that declined substantially when people could interact multimodally. These difficulties also mirrored a strong user preference to interact multimodally. The error-proneness and unacceptability of speech-only input to maps was traced to people's difficulty articulating spatially oriented descriptions. Analyses also indicated that map displays can be structured to minimize performance errors and disfluencies effectively. Implications of this research are discussed for the design of high-performance multimodal interfaces for future map systems.