Research in intelligent agents and multiagent systems that run on the Internet has received increased attention and importance in recent years. Since the Internet continues to grow, intelligent agent technology is progressively being introduced to many Internet-based applications for communications between different applications. The aim of this paper is to present an Internet-based architecture for multiagent systems which offers a communication infrastructure and coordination services for agents to achieve their goals. A structured architecture is proposed to support communication facilities among several agents and coordinate agent activities in distributed environments such as the Internet and Intranets. The architecture consists of 1) Application agents, 2) Communication handler, 3) Knowledge manager, and 4) Repository. (Yuen, et al. 1999; Leung, et al. 1999). These four layers cooperate together and provide common facilities necessary for typical multiagent systems or agent-based applications with various choices. An Internet-based prototype for auditing and detecting unauthorized transactions within an organization over the Internet or an Intranet is implemented to demonstrate the practicability and feasibility of the proposed Internet-based architecture for multiagent systems.