Drying operations are common in food industries. One of the main components in a drying system is the furnace. The furnace operation involves heat-mass transfer and combustion, thus it demands a complex mathematic representation. Since autoregressive methods are simple, and help to simulate rapidly a system, we model a drying furnace of olive pomace via an auto-regression with exogenous variables (ARXs) method. A neural network of radial basic functions (RBFs) defines the ARX experimental relation between the amounts of dry pomace (moisture content of 15%) used like fuel and the temperature of outlet gases. A real industrial furnace is studied to validate the proposed model, which can help to control the drying process. (C) 2012 Elsevier Ltd. All rights reserved.