Multimodal mobile dialog systems face new challenges which do not exist in traditional spoken dialog systems. A mobile dialog system should provide robust task management during network loss, and intelligent assistance in new environments. To provide these capabilities, a system must sense changes in the user environment, and communicate them effectively to the user. In the field of human-computer interaction, such systems are referred to as context-aware systems. Two possible approaches have been explored: in one, the user requests information when needed (information pull), and in the other, the system automatically offers relevant information (information push). In this paper, we present our dialog management strategy for context-awareness using an information push model. A preliminary evaluation conducted in both labaratory and driving environments shows that context-awareness significantly improved user performance.