In this work, an integrated Large Eddy Simulation (LES) model is developed for sooting turbulent nonpremixed flames and validated in a laboratory scale flame. The integrated approach leverages state-of-the-art developments in both soot modeling and turbulent combustion modeling and gives special consideration to the small-scale interactions between turbulence, soot, and chemistry. The oxidation of the fuel and the formation of gas-phase soot precursors is described by the Flamelet/Progress Variable model, which has been previously extended to account for radiation losses. However, previous DNS studies have shown that Polycyclic Aromatic Hydrocarbons (PAH), the immediate precursors of soot particles, exhibit significant unsteady effects due to relatively slow chemistry. To model these unsteady effects, a transport equation is solved for a lumped PAH species. In addition, due to the removal of PAH from the gas-phase, alternative definitions of the mixture fraction, progress variable, and enthalpy are proposed. The evolution of the soot population is modeled with the Hybrid Method of Moments (HMOM), an efficient statistical model requiring the solution of only a few transport equations describing statistics of the soot population. The filtered source terms in these equations that describe the various formation, growth, and destruction processes are closed with a recently developed presumed subfilter PDF approach that accounts for the high spatial intermittency of soot. The integrated LES model is validated in a piloted natural gas turbulent jet diffusion flame and is shown to predict the magnitude of the maximum soot volume fraction in the flame relatively accurately, although the maximum soot volume fraction is shown to be rather sensitive to the subfilter scalar dissipation rate model. (C) 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.