Besides the static soft tissue prediction in surgery simulation, the estimation of postoperative facial mimics is another important criterion for the evaluation of surgery planning. Especially in craniofacial surgery, the realistic prediction of patient's dynamic appearance is of utmost importance. For a realistic simulation of facial mimics at least an adequate biomechanical model of soft tissue including the mimic musculature is needed. In this paper, we present a new shape-based heuristic approach for the modeling of arbitrarily shaped muscles from tomographic data, and the estimation of basic individual facial mimics.