Short-term load forecasting plays an important role in planning and operation of power system. The accuracy of this forecasted value is necessary for economically efficient operation and also for effective control. This paper describes a method of autoregressive (AR) modified covariance (MCOV) in solving one week ahead of short term load forecasting. The proposed method is tested based from historical load data of Malaysia and New South Wales, Australia. The accuracy of proposed method is obtained and reported.