The paper describes a hybrid approach to combustor emissions prediction in which a stochastic model of the multi-scalar field necessary for NOx and smoke prediction is combined with that of a flowfield from a coarse CFD calculation in a representative tubular combustor. A transport equation for the joint composition pdf, for example that of mixture fraction, soot mass concentration and number density, is computed using a Monte Carlo simulation in which the mean velocity and turbulence properties are generated by the CFD calculation. By adopting a standard k-ε / presumed scalar pdf approach for the latter, the CFD and Monte Carlo calculations are completely decoupled and the stochastic modelling of the reacting scalars is effected within readily accessible computing resources.