Our work is related to model-driven Web Services performance evaluation. It aims to integrate a performance analysis into the development cycle of the Web Services by using a model-based approach that generates from the service model a simulation code that allows to simulate the behavior of the Web services before the development is completed. Our approach allows to guess the behavior of a web service even when service code is not available and there is no a priori knowledge about its performance. The performance estimation we obtain can be used to negotiate and evaluate SLAs of the performance of the service. Our work will also be applicable to service compositions according to selected composition patterns. This presentation will show some results we obtained.