Chronic Kidney Disease prediction is one of the most important issues in medical decision making. The discovery of ckd prediction is an important task because it depends on experts of doctor knowledge. Construct effective ckd prediction in time is essential to prevent healthy patients. Chronic kidney disease is one of the leading cause of death and early prediction of chronic kidney disease is important. Prediction is most interesting and challenging tasks in day to life. Data mining play a essential role for prediction of medical dataset. It extract unknown information from hidden knowledge. This paper can proposed a new chronic kidney disease dataset with three classifiers such as radial basis function network, multilayer perceptron, and logistic regression. The obtained result of this experiment shows in terms of prediction accuracy, type I error, type II error, type I error rate, type II error rate, sensitivity, specificity, F-score. Kappa value represents that measure of agreement between the classification made by the experts and classifiers. Accuracy of the three classifiers are evaluated for the new CDK dataset from UCI repository. Thus, the paper discussed the result of comparative study of classifiers in medical ckd dataset.