Coals may be blended at mines to meet contractual criteria set by the client. In formulating a blend, the criterion that is hardest to meet is often the handlability because it is difficult to predict. Most research on coal handling treats coal handlability as a property to be measured for a given material, and there appear to have been very few attempts to predict the consequences of blending choices, or to use such information to optimise coal blends for the different contractual requirements. This paper presents what is believed to be the first study to explore the causal relationships in coal blending, to use these to predict the handlability of a blend, and to establish the basis needed to optimise blending choices. The ultimate goal of this work is an algorithm that can be used in on-line real time choices in coal blending operations, either at mines or at power stations. (c) 2005 Elsevier Ltd. All rights reserved.