Software agent technology has attracted much attention for developing various distributed systems, composed of autonomous agents interacting with one another using particular mechanisms and protocols. Such systems provide high-level reconfigurability, flexibility and robustness in dynamic environments, and have applications in many areas. However, the great advantages of distributed agent systems are often overshadowed by the challenges of providing flexible and consistent security management for agent interactions in the dynamic and heterogeneous computing environments. Trust, which is the belief of an agent that the other agent will act or intend to act beneficially, is a basis for secure distributed agent systems. In this paper, various major research issues of managing trust among various entities are identified, and the approaches in dealing with them are discussed. A framework for managing trust among various entities in distributed agent systems is presented.