An active control system based on statically trained, feed-forward, multilayer-perceptron neural networks was designed and demonstrated, by experiment and simulation, for NO and CO2 from a two-stage laboratory combustor operated under staged-air conditions. The neural networks are arranged in two clusters for feed-forward/feedback control. The first cluster is a neural-network-based, model-predictive controller (NMPC) and is used to identify the process disturbance and adjust the manipulated variables. The second cluster is a neural-network-based Smith time-delay compensator (NSTC) and is used to reduce the impact of the long sampling/analysis lags in the process. NMPC and NSTC are efficiently simple in terms of the network structure and training algorithm. The controller based on NMPC/NSTC showed a superior performance over the conventional proportional integral derivative controller. The novel controller has also been demonstrated on a neural-network-based combustor process simulator.