A two-step process was adopted to model turbulent ignition that takes advantage of the possibility of decoupling the mechanical flow from chemical reaction due to the small amount of heat release before ignition. In the first step, a Reynolds stress model is employed to calculate a chemically frozen, turbulent counterflow. The second step models the ignition event by solving a joint scalar PDF equation using a Monte Carlo technique. The frozen velocity field is used to initialize the PDF model and to govern its evolution. As observed in previous DNS calculations, ignition occurs at a "most reactive" mixture fraction. The present calculations indicated that turbulence intensity had little effect on ignition temperatures, which were about 30 K higher than, but parallel to, laminar ignition temperatures. Similar results were found for both the IEM and modified Curl's mixing model. Turbulent ignition temperatures were similar to laminar ones when the mixing model was modified to account for preferential diffusion. These results are different from turbulent ignition experiments since the experiments did indicate a turbulent intensity effect on ignition of up to 35 K. These discrepancies were attributed to shortcomings in the molecular mixing models in the flows of interest where the turbulent Reynolds numbers are low. A potential source of this problem was identified as the representation of the scalar mixing frequency as a constant ratio of the scalar to flow time. (C) 2003 The Combustion Institute. All rights reserved.