Rotating machines are essential assets in many industries, and critical to the operation of these machines is the health of the rolling element bearings used to support shafts and gears. Condition based maintenance programs allow the health of machine components to be determined, and repairs scheduled at the optimum time, and prior to unexpected failure. One of the most common methods for detecting rolling element bearing faults is vibration analysis, with a number of different techniques available. This analysis compares the fault detection ability of a spectral kurtosis optimized band – pass filter analysis technique with an energy level optimized wavelet decomposition analysis, and presents a basic semi – automated process for diagnosis. Wavelet analysis proved superior in its ability to detect both localized faults and extended outer race faults, whilst band – pass filtering was limited by its lack of time–frequency resolution. The semi–automated process utilized wavelet analysis and proved successful in detecting localized bearing faults.