There are different statistical fault detection indicators applied in the time domain to detect crack propagation in the gear tooth root. ‘TALAF’ and ‘THIKAT’ are two newly presented indicators which have been designed and rec-ommended to improve the performance of ball bearing fault detection after a cer-tain stage of degradation. This paper studies the performance of these two new in-dicators, together with the RMS, kurtosis and crest factor indicators, in the context of detecting faults in the gear tooth root. The paper also presents an investigation of the performance of these indicators in the presence of three levels of random background noise. Gear mesh stiffness calculations and dynamic simulation have been performed using Matlab™ to obtain the residual gear centre point displace-ment signals for different crack propagation cases. The simulations indicate that the RMS and kurtosis perform well for crack depths up to approximately 50% of the tooth root thickness. Kurtosis and THIKAT show the most sensitive perfor-mance with an increasing noise level.