An empirical model for the selective non-catalytic reduction (SNR) of NO with NH3 as reductant is proposed. The model is based on two empirical reactions: one reaction describing the oxidation of NH3 to NO, and the other the reduction of NO by reaction with NH3 to N2. It is shown that the model describes the influence of the major parameters that can be varied in the SNR process (injection temperature, temperature gradient, residence time, amount of injected NH3 and amout of injected additive). The model predictions are compared to experiments carried out in a fluidized bed boiler and in a pilot plant using flue gas withdrawn from a coal fired utility boiler and it is compared with the kinetic model of Miller and Bowman (1989). When ideal mixing between the injected NH3 and the flue gas can be assumed, as in the cyclone of the fluidized bed boiler, a good agreement between the model predictions and the experiments is obtained. The empirical model gives a better description of the result for a fluidized bed boiler than the kinetic model of Miller and Bowman (1989). In the coal fired pilot plant the mixing process is not ideal, this seems to influence the process performance.