This paper presents the application of wavelet transform for broken rotor bar fault detection in a three-phase induction motor. One phase stator current signals are decomposed with db8 mother wave into approximate and detail coefficients. The detail coefficient including the frequencies near fundamental is chosen for analysis. Using the statistical analysis of the detail coefficient, it is found that mean absolute deviation varies in proportion to the fault severity. With the help of this parameter, not only fault detection can be accomplished, but also fault severity can be measured. The performance of the parameter is also tested under significant reduction in voltage drop and it is found that the mean absolute deviation remains undeterred in detecting and identifying fault severity under variation of supply voltage. The study has been performed on the experimentally collected data on laboratory simulated breakages in bars of the rotor. © 2019 IEEE.