Electrical signal analysis(ESA) has been in use for quite some time in detecting faults occurring in induction motors. In industrial applications, the induction motors are always coupled to dynamic loads such as pumps, blowers, compressors, etc. Failure to either the motor or the dynamic load is critical to the operation of the plant and results in loss of production and unscheduled downtime leading to increased costs. Hence there is a need for a cost effective early failure detection system not only for induction motors but also for the dynamic loads connected to the motor. In this paper, an experimentally demonstrated model-based sensorless approach to detect varying levels of cavitation in centrifugal pumps is presented. The proposed approach is sensorless in the sense that no mechanical sensors are used in pump monitoring, in order to avoid any reduction in overall system reliability and increased costs. Rather on-set of pump cavitation is detected using only the motor electrical signals. The developed fault detection scheme is insensitive to electric power supply /mechanical load variations and has been tested on data collected from a centrifugal pump connected to a 3Î¦, 3 hp induction motor. In addition to the staged faults, experiments are also conducted to test the false alarm performance of the algorithm. Â© 2010 COMADEM International.