Screening experiments were carried out to study the reduction of NO emissions from a 20 MW circulating fluidized bed (CFB) boiler, equipped with an installation for selective non-catalytic reduction (SNR). The influence of both primary measures and SNR were evaluated, using 25-1 and 25-2 fractional factorial designs for the two fuels, crushed peat and wood waste, respectively. Polynomial models were deducted from statistical analysis of the experiments, and a good agreement between models and measured data was obtained. The evaluation showed that, by using a designed experimental procedure, CFB operating conditions yielding an NO reduction of 60-80% could be identified, with both primary measures and the SNR being of approximately equal importance. Most important factors for the NO reduction were air:fuel ratio, the amount of NH3 added, the load and the fraction of lower secondary air; but the reduction is also influenced by small interaction effects. A discussion of the use of experimental designs for increased understanding and optimization of combustion processes is also given.