Aiming at the problem of extraction difficulty of early non-stationary weak fault signal feature,low resolution of characteristic parameter,early fault diagnosis difficult exist in the rolling bearing fault diagnosis,a fault diagnosis method based on least squares mapping（LSM） fault characteristic parameter optimization and support vector machine（SVM） is proposed.Firstly,the non-dimensional symptom parameters（NSPs） in the time domain can reflect the features of the vibration signals measured in each state are calculated.Then,the high sensitivity symptom parameters are built by optimizing the calculated non-dimensional symptom parameters（NSPs） in the time domain with the LSM theory.Finally,the symptom parameters selecting by sensitivity identification factor are input to the SVM for the fault diagnosis,the fault type of bearing is identified through the sequential inference diagnosis.The practical examples of fault diagnosis for a motor bearing are shown to verify that the method is effective.