Modeling of partially stirred reactors (PaSR) by stochastic Monto Carlo simulations is carried out to investigate the relative performance of three existing turbulent mixing models: the interaction by exchange with the mean model, the modified Curl mixing model, and the Euclidean minimum spanning tree model. A detailed mechanism for hydrogen oxidation, which involves 9 species and 19 reactions, is incorporated into the simulations using the in situ adaptive tabulation algorithm. Numerical simulations are performed for a wide range of residence and mixing times revealing the significant differences in the PDFs of mixture fraction, in the scatter plots, and in the extinction behaviors between the three different mixing models. The conditional mean scalar dissipation implied by each model is deduced analytically or numerically in the PaSR, but it is shown not to have the same significance as in the conditional moment closure and flamelet models. (C) 2003 The Combustion Institute. Published by Elsevier Inc. All rights reserved.