This paper aims to present the performance of a simple nonlinear time domain transformation known as Curve Length Transform (CLT) on the detection of rolling element bearing faults. Experimental vibration velocity and acceleration responses collected from a test rig are used to analyze the bearing signals. Artificial defects are introduced on the raceways of inner and outer rings of a deep groove ball bearing to examine different bearing faults. Unbalance load condition is created by attaching an eccentric mass to the rotating shaft in order to analyze the effect of the unbalance load together with the bearing fault on the performance of the curve length transform. Basic temporal indicators such as Root Mean Square (RMS), standard deviation, peak to peak, crest factor and kurtosis derived from the curve length signals are used for defect detection purpose. Time domain analyses based on the experimental measurements show that the curve length transform can be used to enhance the diagnostic capabilities of some time domain indicators.