In this paper, we propose a system for context aware battery management that warns the user when it detects that the phone battery can run out before the next charging opportunity is encountered. At the heart of this system, are algorithms that predict: (1) when the next charging opportunity will be available, (2) how much call-time will be required by the user in the interim, and (3) how long the battery will last if the current set of applications continue to execute. We propose algorithms that process user’s location traces and call-logs for making some of these predictions. We also propose a technique to predict battery consumption of applications. We present the design of the system and demonstrate its feasibility by experimentally showing that each of the prediction algorithms can perform with fairly high accuracy.