Locomotive rolling bearing is one of the most important aspects of locomotive fault detection. HHT is a new type time frequency analytical method with adaptability, which can effectively analysis non-stationary signal produced by vibration of locomotive rolling bearings. According to the typical fault produced in locomotive rolling bearing, a system of locomotive rolling bearings fault diagnosis was developed, which was based on HHT, Lab VIEW and PXIe. This system can extract and analysis the fault feature information of locomotive rolling bearings successfully and then identify the working condition and fault type of locomotive rolling bearings accurately. From the experiment results, failure frequency which is obtained by HHT matches theoretical value. It is a effective method to diagnose rolling bearings.