Predictions of a two-dimensional, axisymmetric combustion model, using various devolatilization submodel options, are compared with new experimental data from a near-laminar, drop-tube furnace. Included in the devolatilization submodels that were tested are the commonly used empirical one- and two-step models and a chemical, coal network model with parameters based on coal structure. The goals of this work were to evaluate the latter approach as compared with the simple, empirical approach usually used-in such calculations and to assess the role of turbulence in a near-laminar reacting flow. Comparisons were made for carbon conversion, radially averaged oxygen and near-effluent NOx concentrations, for a range of coal types and equivalence ratios. The predictions quantify an ignition delay which is consistent with the measurements. Computations with the fundamental, chemical devolatilization submodel gave superior predictions of mass loss when the coal type was within the interpolation range of the submodel parameter database. Accuracy declined significantly when the coal type was outside the interpolation range. Inclusion of the effects of turbulence was required to account for the observations. Near-effluent NO predictions with the chemical submodel agreed with measured NOx values to within an average of about 20 percent.