In the present study the detection of the crack in rotor system has been carried out using the statistical features such as kurtosis, skewness and crest factor of vibration response. Such features, mainly kurtosis have been used in literature for identifying bearing and gearbox defects. The simulation results of crack detection have been compared with those of experiments carried out using laser vibrometer. Single and two cracks have been identified. It has been noted that the kurtosis and skewness changes are significant when the crack is closer to the bearing locations. The skewness is mostly negative but as the crack depth increases the positive values of skewness increase beyond the middle part of the shaft. The crest factor deviation curve is developed, which could successfully identify even cracks of small depths. © KRISHTEL eMAGING SOLUTIONS PRIVATE LIMITED.