Most works proposed in the field of Computer-Assisted Surgery focus on orthopedics, with anatomical structures (the bones) that are considered as rigid. As the human body is mainly composed of biological soft tissues, deformable biomechanical models have been introduced to assist the surgical gesture with the prediction of tissues deformations. This talk presents the principles that are chosen by our group for the biomechanical modeling of living tissues. Our methodology aims at designing Finite Element models (1) that follow classical biomechanical constraints (i.e. hexahedral elements embedded in a topology that allows the identification of distinctive structures), (2) that can be automatically generated to match each patient morphology, and (3) that are implemented by the mean of classical and validated Finite Element packages. This methodology is illustrated with three clinical applications, namely plastic and maxillo-facial surgery, orbital pathology (the exophtalmia reduction), and sleep apnea syndrome.