The relationships between coal properties, the strength of coke, and the performance of a blast furnace were investigated by means of correlation studies, and a linear programming model was developed for the prediction of a coal blend based on these relationships. The unique feature of this approach is that it utilizes a knowledge of the minimum blast-furnace coke requirements, rather than the maximum attainable coke quality. The model predicts the least expensive coal blend that would still comply with the minimum coke-quality requirements of a blast furnace. The study also confirms the existence of a relationship between the rank and caking parameters of coal, and the hot strength of coke. The effect of coke-oven operating parameters on coke quality is also discussed.