Minimally invasive surgery is common in present hospitals since it has many advantages over traditional open surgery. However, the skills which it requires are very different from those developed in doing traditional surgery, and they have to be learnt in a different way. A survey of the current training methods reveals that using virtual reality simulators is a promising direction for training minimally invasive surgery, and it requires the developers to improve accuracy and efficiency of human tissue simulations. A further survey of existing deformable models for real time surgical-training simulations leads to the conclusion that there is room for research for developing new thin tissue models in which one of the dimensions is significantly smaller than the other two. A new generic model is proposed for deformation and cutting of thin deformable tissues in real time. The tissues may develop wrinkles when pressed with virtual probes, as well as they can be trimmed and cut with different virtual surgical instruments using desktop haptic devices. The model is then applied to three tissues for simulating virtual arthroscopic surgery: meniscus, ligament and cartilage. The main attention was paid to the meniscus to be realistically deformed while developing wrinkles when tested with the virtual probe, as well as to be cut and trimmed at its edge with various virtual instruments. Ligaments were simulated for displaying their realistic deformation when examined with the surgical hook and cartilage was IV mostly used to modeling its geometry and elastic properties when examined with haptic devices. A side-by-side comparison with actual surgical videos verifies the simulation accuracy. To demonstrate its flexibility, the model was also applied to other objects such as a piece of meat and a thin paper sheet. To validate the obtained results, a virtual reality knee arthroscopic prototype simulator is developed. The proposed models are combined with a scene graph structure and implemented using SOFA. The performances for the implementation of different settings of collision meshes are compared and discussed. The usability and feasibility of the proposed methods are further verified by the real time simulations of various surgical procedures, which include meniscus examination, punching out torn tissues, removing tissues pieces by pieces, and contouring the edge of meniscus.