Mobile agent technology is an evolving paradigm that combines the inherent characteristics of intelligent agents, namely, adaptability, reactivity and autonomy with mobility. These characteristics of mobile agents provide an excellent means of meeting the distributed and heterogeneous requirements for many electronic commerce applications involving low bandwidth and intermittently connected networks. However, the lack of security in the form of code confidentiality renders this paradigm unsuitable for commercial software. In this paper, we address the problem of mobile agent security by proposing a novel method of mobile agent obfuscation using the concept of opaque predicates to prevent adversaries from observing the control flow of agent code. We discuss about the efficiency of our proposed methodology by demonstrating that to an adversary, the problem of determining the outcome of such opaque predicates is often intractable.