In this study, an effective method based on wavelet transform, for identification of damage on rotating shafts is proposed. The nodal displacement data of damaged rotor is processed to obtain wavelet coefficients to detect, localise and quantify damage severity. Because the wavelet coefficients are calculated with various scaled indices, local disturbances in the mode shape data can be found out in the finer scales that are positioned at local disturbances. In the present work the displacement data are extracted from the MATLAB model at a particular speed. Damage is represented as reduction in diameter of the shaft. The difference vectors between damaged and undamaged shafts are used as input vectors for wavelet analysis. The measure of damage severity is estimated using a parameter formulated from the distribution of wavelet coefficients with respect to the scales. Diagnosis results for different damage cases such as single and multiple damages are presented.