The load forecasting aims at the energy management in the field of power supply systems. It helps to diminish the production cost, spinning reserve capacity and enhance the reliability of the power system. It is tremendously essential for financial institutions, electric utilities and other participants in electric energy market, be it for transmission, generation or distribution. The economic allotment of electricity generation plays a vital role in short term load forecasting. This paper presents a solution methodology based on Levenberg Marquardt algorithm of an artificial neural network technique for short term load forecasting. The system data for forecasting the load includes the parameters like dry-bulb temperature, dew point temperature, humidity and load data. The live load data was recorded from the 66kV substation located at Bhai Roopa, Bathinda in Punjab state of India. The corresponding weather data was collected from the Indian Meteorological Department "IMD" at Pune in Maharashtra state for the years 2015 and 2016. The Levenberg Marquardt algorithm had been implemented to minimize the error function derived on the basis of computed load and actual load. This work had been carried out using the MATLAB software. The obtained results would support an effective and accurate load forecasting in future.