This thesis deals with the fault management of concurrent and distributed applications. It explores how they can be made easier to manage, and - as a long term goal - how they can become self-managing. The goal of fault management is to detect, diagnose and repair faults that may occur in the application at run-time. This thesis searches for management solutions that are more effective than those currently existing, and have potential for being reused and automated. It proposes a novel approach that overcomes many of the difficulties encountered currently in fault management. It constitutes seminal work that outlines the problems and possible solutions and validates them through selective prototyping. Obviously, it does not yet provide a full-fledged and turnkey solution to automatic management.