At the present time, computer models for coal combustion are not sufficiently accurate to enable the design of combustion plant or the selection of a coal based on combustion behaviour. Most comprehensive combustion models can predict with reasonable accuracy flow fields and heat transfer, but usually with a much lesser degree of accuracy than the combustion of the coal particles through to char burnout. Many research programmes are aimed at developing a much more accurate predictive tool for assessing coals specially fired in burners or furnaces employing a range of NOx abatement technologies. Some of the current developments in CFD coal combustion modelling are outlined here, Particular attention is paid to the first step, where the devolatilisation pre-processor code is used to compute the pyrolysis rate, the yields and the composition of volatiles and char. These parameters are used as inputs to the devolatilisation and volatile combustion sub-models, where various options can be used, and also the char burnout sub-models. The accuracy of the sub-models is examined using data from four well-studied coals, three from the UK and one from the US. The main network devolatilisation codes are compared with experimental data. Two char combustion models have also been investigated in order to compare char burnout predictions and the development of char morphology and surface area during burnout are considered. The applications of these sub-models to two combustion situations were considered. These involve reactions in a drop tube furnace and a low NOx industrial burner and in both cases, the model predictions were compared with experimental measurements.