Abstract: In order to solve the problem of early weak fault diagnosis of bearings, the intrinsic time-scale decomposition (ITD) is proposed. ITD and cuckoo adaptive stochastic resonance (CASR) are combined to extract feature frequencies of early weak faults of rolling bearings. Rolling bearing vibration signals were collected for complex and low signal-to-noise ratio (SNR) and difficult problem to extract the fault features, the combination of ITD can inhibit the endpoint effects and low computation complexity, this method will take signal-to-noise ratio as the objective function of the stochastic resonance, through the simulation signal analysis and examples, will adopt ITD as pretreatment of CASR signals processing. The signal-to-noise ratio can be increased by 2.17 times, and the fault features can be effectively extracted.