Kidney transplantation is one of the effective post-dialysis treatment methods for patients with chronic renal failure in the world. Most medical data are imbalanced and the output of algorithms is inefficient with imbalanced data. The aim of this study is to predict the two-year survival rate of kidney transplant patients and provide a more accurate model. We evaluate the data of renal transplant patients in Afzalipour Medical Education Center 2006-2010, Kerman, Iran. Survival prediction of kidney transplantation with MLP and RBF neural networks with two methods of sampling and investigating the factors affecting the survival of kidney transplant in renal transplant patients is considered by the binary particle optimization algorithm and nearest neighbor algorithm. Accuracy of the results can be increased by using the oversampling method in imbalanced medical data, and radial base network model is a suitable model for predicting the survival of kidney transplant patients.