Many agent systems have been developed and suggested for commercial application. However, in spite of the significant potential offered by the agent paradigm, the lack of such important properties as security, anonymity and untracebility especially in open dynamical environment, such as the Internet, has blocked the active implementation of agent technologies. Protecting agent systems poses a more demanding challenge comparing to traditional environments. Apart of all the threats which face the latter, there are attacks specific to agent environments. The chapter considers issues of agent security, anonymity and untracebility and outlines the main threats to agent systems and existing protection methods.