In this article the probabilistic composition of criteria is applied to the evaluation of graduate courses in the area of Production Engineering (Industrial Engineering in some regions). The difference between analyses based only on outputs and those that counterbalance output measurements with measurements of the volume of faculty resourcees applied is investigated. Indicators of the evolution through time are also constructed. The basic output variables measure formation of M. Sc. and publication of research findings. Adding other variables does not substantially change the conclusions. It is demonstrated that, in the data set studied, the probabilistic approach offers an objective tool for the global evaluation of course performance and, especially, of the evolution of such performance through time.