In recent years,the rapid development of machine learning has made unprecedented achievements in the fields of computer vision,natural language processing and data mining. There are many disciplines of earthquake research,including seismology,geodesy,geochemistry,geomagnetism and geology. Therefore,the multi-source,complex and massive data generated by the earthquake research highly meet the requirements of machine learning for training data. In fact,many scholars have applied machine learning methods into earthquake prediction. In the view of the background of machine learning,the application process and evaluation methods of earthquake prediction,in this paper we reviewed the application and main progress of earthquake prediction based on machine learning methods and data from different disciplines in recent years,and summarized and discussed the application of machine learning in earthquake prediction.