Coal blends are commonly used in pulverised fuel fired power plants. Past experience has shown that some coals exhibit synergistic effects when co-fired with other coals. Despite the recent progress in the field, there is still a general lack of understanding about how and why such synergistic effects take place. It is, therefore, imperative to develop reliable techniques to predict the combustion characteristics of coal blends. In the present paper, a commercially available computational fluid dynamics (CFD) software package, known as Fluent, is applied to simulate the combustion of binary coal blends of Australian black coals in a pilot-scale furnace. The modelling was mainly concerned with employing the two-mixture-fraction-pdf approach to separately track the individual components of each blend. Additionally, another approach, which takes the blend as a single coal with the weighted average properties of the components, was also used and compared. The main combustion characteristics of coal blends, such as ignition, burnout and NOx emissions were studied and corresponding predictions were validated against measurements in a pilot-scale furnace. The comparisons indicated that the CFD model with the two mixture fraction approach can successfully predict the non-additive (synergistic) combustion behaviour of coal blends, providing an effective tool for full-scale applications. The single coal approach presents nearly additive prediction of coal bends, and is only applicable to the blends of similar coals.