To estimate the temperature profiles in a graln-flar-bed heat conduction process, it is important to know tile thcrinal properties accurately. In this paper, online parameter identification of the thermal properties was carried out by using the Extended Kalman Filter. A state space model was derived from the mathematical model of the grAnUlar-bed process by use of finite integral transformation, and the Extended Kalman Filter was used to estimate state variables and parameters simultaneously. The proposed parameter-identification method was applied experimentally to a laboratory scale granular-bed process. By using a small number of temperature measuremonts as observed variables, the Extended Kalman Filter provided accurate parameter identification even in the presence of measurement and system noises.