Abstract: Power system load forecasting using stochastic system state model identification technique is proposed.First,a power system load model is presented with a relation analysis method for determining its order and estimating its parameters.Then Kalman Filter theory is used to obtain one-day-ahead load forecasting in various period.In this paper,the data used for modelling,forecasting and error analysis are real load values from the north-east power system during the period of 1982-1984,calculation was performed on PDP-11/23.