VDL Weweler is a producer of air suspension systems. The base of these systems consists of spring steel. In order to reshape this spring steel component, it is heated in a walking beam furnace. Conventional control systems of such furnaces maintain the furnace temperature at target, while the furnace speed is fixed. These control strategies can function well under steady operations (continuous production at a fixed furnace speed), but they are less suitable in transients which arise after startup and stops. In practice, a furnace rarely operates in complete steady state conditions. In those transient situations, the furnace has to be controlled manually, since there is no information on the product temperature. This paper describes and evaluates the application of the DotX Nonlinear Predictive Controller (DNPC) to the VDL Weweler walking beam furnace. This controller belongs to the Nonlinear Model Predictive Controller (NMPC) family, in which control actions rely on on-line optimized model predictions. The model, used in DNPC, accurately predicts the temperature of all products in the furnace, depending on their geometry and the furnace’s speed and temperature. By adjusting the set points for furnace temperature and speed, DNPC is able to keep the product temperatures on target, while energy consumption is minimized, even during large transients. The DNPC controller runs on an external CPU which communicates with the furnace via an OPC server. Initial tests have shown that controlling the walking beam furnace at VDL Weweler by DNPC results in a reduction of CO2 and energy consumption of up to 10 % compared to the conventional furnace controller. Furthermore, the product temperature is maintained automatically at target by DNPC, even at startup and during transients. Moreover, the target product temperature can be changed online and is followed automatically.