Abstract-The increasing pervasiveness of web browsing on portable devices calls for the careful consideration of how web pages are accessed by mobile users. While flat-rate pricing (independent of the usage) has been the conventional data pricing model in rich countries, mobile roaming users and users in developing regions are typically subject to usage-based pricing based on cost per megabyte downloaded. With the exceptional growth in both the size and complexity of web pages, usage-based pricing can result in high monthly bills even under nominal web usage.In this paper, we propose a cost-aware mobile web browsing mechanism that aims to substantially reduce the usage cost incurred for the end-user due to usage-based pricing. The key idea of our approach is to leverage the data plan of the user to compute a cost quota for each web request and a network middle-box to automatically adapt any web page to the cost quota. Given the complexity of existing web pages, we leverage a suite of content adaptation techniques to automatically produce the best possible version of a web page for a given quota. Based on a detailed analysis of popular web pages, we show that our system can significantly reduce data costs while extracting the best possible version of a page given the cost constraints.