Bearings are precision made components that enable machinery to move at extremely high speeds with low friction and also support loads acting on the shaft. It is vital to identify an early fault in bearing to avoid catastrophic damages. In this work condition monitoring model is developed and it comprise of TQWT (Tunable Q-factor wavelet transform) which is based on decomposing a non-stationary vibration signal into sub-bands, Spectral features were applied to the sub-bands of TQWT for feature extraction and different classification techniques were used for classify the various conditions of bearing. © 2017 Elsevier Ltd.