Stroke is the second leading cause of death with 11.13% of total deaths worldwide. Until now there has been no effective effort to combat stroke, both by increasing public awareness and optimal stroke management. Certainty to determine the type of stroke early is very important to prevent the danger of stroke. Determination of the type of stroke can be done by analyzing the clinical data of stroke patients. In addition, the application of data mining using the Naive Bayes Classifier method is expected to be able to identify the type of stroke. Sample data were taken from the medical records of stroke patients at a hospital in Central Java, Indonesia between 2015 and 2016, with a sample of 738 data. The patient's clinical data consists of 25 diagnostic criteria attributes which contain the results of a physical examination, patient symptoms, medical history, and laboratory tests. The output of this system is the prediction of the patient's stroke type: thrombotic stroke, embolic stroke, systemic hypoperfusion, intracerebral hemorrhage, and subarachnoid hemorrhage. The results of this study have an accuracy rate of 84.61%. This application is expected to facilitate neurologists in determining the type of stroke in patients.