This paper presents a soft-sensor for on-line monitoring of fuel moisture in a furnace. The method utilizes combustion power estimation and a dynamic model of the secondary superheater. In addition, the time delays in detecting changes in moisture content of the fuel are small enough for the method to be used for controlling air and fuel feed and preventing steam and pressure oscillations. To verify the fuel moisture soft-sensor, experiments were performed at a BioPower 5 CHP plant, which utilizes BioGrate combustion technology for very wet biomass fuels with a moisture content as high as 65%. Finally, the results are analyzed and discussed. (C) 2013 Elsevier Ltd. All rights reserved.