In this paper, a novel method based on former cases for plastic surgery prediction is presented. This method takes a pre-operative frontal facial picture as an input. Landmarks of the face are then extracted and constitute a distance vector. As a set of facial parameters, such a vector is entered into either a support vector regression (SVR) predictor or a k-nearest neighbor (KNN) predictor which is trained on a set of pre- and postoperative facial distance vectors of former cases. After the predicted distance vector generated, new landmarks positions are updated and the final result is generated in terms of changes between predicted landmarks and the original ones. Several experiments are carried out and the results show a great accuracy of prediction, which proves that this method is of high validity.