Computational fluid dynamics is a widely used tool in optimizing natural gas burners, for instance, for emission issues. Especially, a further reduction of NOx emissions is of interest. However, due to computational efforts calculating three-dimensional turbulent flames, there is the necessity for simplified models in order to simulate the combustion reactions and the NOx formation, respectively. Hitherto, models describing thermal NO and prompt NO formation, respectively, were applied in a post-processing step. Beforehand, the flow field including combustion has been determined in the three-dimensional geometry. However, in the former work, it was shown that prompt NO formation is of minor significance. For temperatures higher than 1600 degrees C, thermal NO formation is dominating. At lower temperatures, the N2O/NO and NNH route have significant contribution. Though, the widely applied prompt NO model captures the observed trends acceptable, it lacks of physical bases. Besides low temperature NO formation is more related to N2O/NO and NNH route, it assumes the prompt NO formation to be proportional to the fuel concentration. The detailed reaction mechanism show NO formation more related to fuel oxidation rate, i.e. radical concentration. Thus, in this work, a new simplified model combining thermal NO formation, N2O/NO, and NNH route is proposed. It applies steady-state approximation for the intermediate species, i.e. N, N2O, NNH, and NH. In this way, their concentrations can be obtained by four algebraic equations and rate of NO formation can be calculated without any model parameter, solely based on reaction kinetics. Moreover, the concentrations of O-2, N-2, H-2, and H2O as well as the radicals O, H, OH, and HO2 have to be known from combustion calculations. The model was evaluated against the predictions of a detailed reaction mechanism, showing good agreement in a wide range of conditions. Neglecting prompt NO formation affects predicted NO emissions only under very fuel rich conditions. Under these circumstances, total NO formation is low, anyway. Thus, the performance of the presented model is not influenced by the lack of prompt NO formation. (c) 2005 Elsevier Ltd. All rights reserved.