A new method based on support vector machine（SVM）model is proposed to predict settlement of road soft foundation. A case study shows that the prediction results accord well with the actual settlement measured data. The new method is also compared with BP artificial neural network model and traditional hyperbola method. The prediction results indicate that the SVM model has a better prediction ability than BP neural network model at the same training set mean-square error. Utilizing the settlement data under multi-stage loading

SVM model has a better reflection for foundation soil deformation trend compared with hyperbola method only using the data under pre-loading. Therefore

settlement prediction based on SVM model can reflect actual settlement process more correctly.