Liver disease is a liver disease caused by viruses, alcohol, lifestyle and others. Someone often does not realize or is late to know liver disease so that when examined liver disease is severe, it would be better if treatment is done early by knowing the symptoms suffered. Data mining can help diagnose liver disease more easily, especially to help doctors determine whether patients suffer from liver disease or not, with symptoms almost close to liver disease. The process of diagnosing liver disease is carried out by a classification process and the result is that the patient suffers from liver or not. This research uses a data mining classification method using an artificial neural network method, namely Multilayer Perceptron. The Indian Liver Patient Dataset (ILPD) used in this study was obtained from the UCI Machine Learning Repository. The division of the data set over the training data and test data is done by Cross Validation. Performance measurement of the method uses confusion matrix. Based on the research conducted, it was found that the application of Multilayer Perceptron resulted in varying accuracy based on testing with different Fold values with the highest accuracy of 83.70% when the Fold was 7, and the lowest accuracy of 80.57% when the Fold was 3. Then the average accuracy of all Fold tests is 82.13%