Tool failure model evaluation and parameter identification are required in order to obtain given goals by implementation of tool replacement policies. The case of multi tool systems, such as e.g. milling cutters, has some interesting features, in that individual cutting edge life is affected inter alia by overall cutter performance. A multi state Markovian model is adopted in order to derive a discrete description of the current condition, and the evolution, of the set of individual cutting edges of the cutter. The number of states required depends upon the number of cutting edges, and the discrete levels describing the conditions observed in the course of service life-(from new to completely failed). Given proper experimental evidence parameter identification is performed with a maximum likelihood algorithm. Evaluation of residual life at any given stage is one of the attractive features of the model for process control application.