The high growth of Multi-Agent Systems (MAS) in Open Networks with initiatives such as Agentcities1 requires development in many different areas such as scalable and secure agent platforms, location services, directory services, and systems management. In our case we have focused our effort on security for agent systems. The driving force of this paper is provide a practical vision of how security mechanisms could be introduced for multi-agent applications. Our case study for this experiment is Carrel [9]: an Agent-based application in the Organ and Tissue transplant domain. The selection of this application is due to its characteristics as a real scenario and use of high-risk data for example, a study of the 21 most visited health-related web sites on the Internet discovered that personal information provided at many of the sites was being inadvertently leaked for unauthorized persons. These factors indicate to us that Carrel would be a suitable environment in order to test existing security safeguards. Furthermore, we believe that the experience gathered will be useful for other MAS. In order to achieve our purpose we describe the design, architecture and implementation of security elements on MAS for the Carrel System.