Turbulent mixing with fast chemical reaction was modeled in the confined jet flow at large Schmidt number (Sc similar to 1000). The Reynolds-Averaged-Navier-Stokes approach was employed to describe the turbulence-chemistry interaction. Numerical results were validated against data obtained by the two-color planar laser-induced fluorescence method. Based on these data the dissipation rates of the mixture fraction and the product reaction concentration were calculated. Their comparison showed essential differences both in dynamics and values. The influence of different-order finite-difference schemes for scalar gradient determination and of the noise-signal ratio on the dissipation rates was estimated. More accurate higher-order schemes caused both noised and corrected dissipation rates to increase. Eliminating the noise leveled essentially the order-effect of finite-difference schemes. The analysis of the mixing models was demonstrated that the mixing model with the constant mechanical-to-scalar time ratio R and the Multi-Time-Scale model overestimated the mixture fraction variance sigma(f) in comparison with experiment. To predict the dissipation in the jet flow at Sc similar to 1000 the low-Reynolds-number effects were considered: in the transport equation for sigma(f) the ratio R and the turbulent Schmidt number Sc-sigma were the functions of turbulent Reynolds number. The accuracy of the used standard k-epsilon model was improved by coordinating the values of the model constants C-mu and C-epsilon 2 with the jet velocity decay and the expansion radius (C-mu=0.06 and C-epsilon 2=1.87). The examination of the known reaction rate models for the gas flow (Sc similar to 1) (the segregation intensity approach, the EDC-model or the PDF method) showed that the reaction rate was overestimated by these models in comparison with the one obtained from the measurements in the liquid flow. The proposed modification of the EDC-model took into account the specific micromixing in vortex structures of flow at Sc >> 1. (C) 2011 Elsevier Ltd. All rights reserved.