Creating a global information-sharing environment in the presence of autonomy and heterogeneity of data sources is a difficult task. When adding mobility and wireless media to this mix, the constraints on bandwidth, connectivity, and resources worsen the problem. Our past research in global information-sharing systems resulted in the design, implementation, and prototype of a search engine, the summary-schemas model, which supports imprecise global accesses to the data sources while preserving local autonomy. We extended the scope of our search engine by incorporating mobile agent technology to alleviate many problems associated with wireless communication. We designed and prototyped a mobile agent-based secure mobile data access system (MAMDAS) framework for information retrieval in large, distributed, and heterogeneous databases. In order to address the mounting concerns for information security, we also proposed a security architecture for MAMDAS. As shown by our experimental study, MAMDAS demonstrates good performance, scalability, portability, and robustness.