The development of reduced chemistry for implementation in a turbulent reacting computational fluid dynamics (CFD) code for simulations of selective noncatalytic reduction (SNCR) performance in boilers and furnaces is discussed in this paper. To obtain realistic estimates of SNCR performance in large boilers and furnaces, it is important to couple together the important physical processes, heat transfer, reagent mixing, and finite rate chemistry in one simulation. Decoupling the mixing from the chemistry or oversimplification of the finite rate chemistry can lead to erroneous predictions of performance. Using an automated strategy, based on conventional reduced mechanism techniques, a 10-species reduced mechanism was developed based on conditions suitable for the application of lean SNCR with NH 3 or urea reagents. Comparisons between the detailed, reduced, and previously used global mechanisms demonstrated excellent agreement between the new 10-species reduced mechanism and the detailed chemistry on which it was based over a range of conditions appropriate for SNCR. The results also showed significant improvement over the previously used global mechanism. The 10-species reduced mechanism also was incorporated into a three-dimensional, turbulent reacting CFD code for simulation of industrial-scale SNCR systems. Comparisons with detailed chemistry in a simple plug flow geometry showed excellent agreement in predictions of NO reduction, NH 3 slip, as well as N 2 O and HNCO emissions. Comparisons between model predictions and data obtained in a pilot-scale furnace also showed good qualitative and quantitative agreement.