A Flexible Machining System (FMS) is aptly described as an automated job-shop. Machines are connected to one another by an automated material transporter and the system is intended to process a number of different kinds of parts. The system incorporates redundancy of function which allows the processing of parts even though machine failures may have occured. This research develops a method for scheduling parts that considers and, in fact, anticipates machine failure. The method relies on a recent technique from Networks of Queues Theory to characterize the behavior of FMS's. It is called Mean-Value Analysis.