A flow updating model is presented in which the flow is estimated using outputs from a physically based watershed model (UBC watershed model) and a feedback of the most recent flow measurement. These outputs (the flow components) are modified by certain parameters that are updated whenever a flow measurement is available. The updating process is based on a state-space model where Kalman filter technique is used to update the parameters from their past values and the most recent flow measurement. The extent of updating is controlled by the relative uncertainties in the flow measurements and the parameters. The updating model has been applied on the Illecilewaet basin and the flow forecasts have shown great improvement over the ones obtained by using the UBC watershed model only. The flow updating model is formulated based on an assumption that the errors in the watershed model output are of a certain linear structure. The validity of such assumption could be tested by comparing some statistical measures of performance. Key words: real-time, flow forecasting, updating, Kalman filter, state-space, linear errors.