Moisture content is one of the key controlled variables in drying processes. However, this variable cannot or is difficult to be measured directly and it is often an inferred quantity based on experience. Therefore, there is a need to design a state observer to estimate the moisture content on-line for the purpose of direct control of drying product quality. A linear state space dynamic model is used to describe drying in continuous fluidized bed dryers. The estimation technique based on Kalman filter design is used to provide state estimates for an optimal state feedback control system. The filter shows acceptable performance in reducing the noises present in the system and in converging to the actual states from incorrect initial states. Also, state feedback controller shows acceptable performance in tracking set points changes when using either actual states or estimated ones. (c) 2004 Elsevier Ltd. All rights reserved.