Massively Multiplayer Online Games (MMOGs; e.g., World of Warcraft), virtual worlds (VW; e.g., Second Life), social networks (e.g., Facebook) strongly demand for more autonomic, security, and trust mechanisms in a way similar to humans do in the real life world. As known, this is a difficult matter because trusting in humans and organizations depends on the perception and experience of each individual, which is difficult to quantify or measure. In fact, these societal environments lack trust mechanisms similar to those involved in humans-to-human interactions. Besides, interactions mediated by compute devices are constantly evolving, requiring trust mechanisms that keep the pace with the developments and assess risk situations. In VW/MMOGs, it is widely recognized that users develop trust relationships from their in-world interactions with others. However, these trust relationships end up not being represented in the data structures (or databases) of such virtual worlds, though they sometimes appear associated to reputation and recommendation systems. In addition, as far as we know, the user is not provided with a personal trust tool to sustain his/her decision making while he/she interacts with other users in the virtual or game world. In order to solve this problem, as well as those mentioned above, we propose herein a formal representation of these personal trust relationships, which are based on avataravatar interactions. The leading idea is to provide each avatar-impersonated player with a personal trust tool that follows a distributed trust model, i.e., the trust data is distributed over the societal network of a given VW/MMOG. Representing, manipulating, and inferring trust from the user/player point of view certainly is a grand challenge. When someone meets an unknown individual, the question is “Can I trust him/her or not?”. It is clear that this requires the user to have access to a representation of trust about others, but, unless we are using an open source VW/MMOG, it is difficult —not to say unfeasible— to get access to such data. Even, in an open source system, a number of users may refuse to pass information about its friends, acquaintances, or others. Putting together its own data and gathered data obtained from others, the avatar-impersonated player should be able to come across a trust result about its current trustee. For the trust assessment method used in this thesis, we use subjective logic operators and graph search algorithms to undertake such trust inference about the trustee. The proposed trust inference system has been validated using a number of OpenSimulator (opensimulator.org) scenarios, which showed an accuracy increase in evaluating trustability of avatars. Summing up, our proposal aims thus to introduce a trust theory for virtual worlds, its trust assessment metrics (e.g., subjective logic) and trust discovery methods (e.g., graph search methods), on an individual basis, rather than based on usual centralized reputation systems. In particular, and unlike other trust discovery methods, our methods run at interactive rates.