A load monitoring system can be used to monitor the power consumption of a load. Traditionally, a measurement device is needed for an individual load. In this paper, a real-time load recognition algorithm is applied to a non-intrusive load monitoring (NILM) system. Data for voltages and currents are acquired through the monitoring device installed in the power entrance. To achieve real-time load recognition, the Self-Organizing Maps (SOM) based on effective input features is used to classify the loads on line. The system performing the NILM is developed on virtual instrument (VI) using Lab VIEW software associated with a data acquisition (DAQ) card. The performance of the system has been tested and verified by using practical combinations of loads. Promising results have been obtained and presented in this paper.