This paper presents a study on the use of time series regression model for forecasting Malaysian electricity demand with various non-deterministic factors influencing demand. The data of electricity demand in this study is provided by Tenaga Nasional Berhad, the main electricity supplier for Malaysia. Factors influencing the load demand include temperatures, holidays, daily and monthly seasonality. The data comprises of daily peak electricity load Lt (megawatts/hour, MWh-1) in Peninsular Malaysia from January 1997 until December 2000. Due to the nature of the data, time series regression model with autoregressive errors where the errors are serially correlated among observations is proposed. This enables the modeling of serially correlated error using Box-Jenkins autoregressive model. Forecast for one month ahead reveal that a time series regression model with load reduction weights yield better accuracy. Model validation is performed by comparing model predictions with the standard Box-Jenkins model. The results obtained bear out the suitability of the adopted methodology for the forecasting short-term electricity load demand.