Abstract: Aiming at the problem of fault diagnosis for rolling bearing of Urban Transit trains, a method combined empirical mode decomposition (EMD) with envelope spectrum was researched on the basis of roller bearing vibration signals. Rolling bearing vibration signals were decomposed into a finite number of intrinsic mode functions(IMFs) by using EMD. Envelope spectrum was used to calculate some IMFs including the main information. Fault pattern was determined by contrast with characteristic defect frequencies of rolling bearing. The experiment result indicated that the fault pattern of rolling bearing could be identified effectively by the researched method.