A mathematical model of coal blending optimization for blast furnace injection is proposed. Because Chinese current blast furnaces blend pulverized coal by simple mix and traditional lab blending optimization needs plenty of work for the larger number of characteristics of pulverized coal. The new model is based on the principal component analysis indexes P1, P2 and P3 combined with the proximate analysis and elementary analysis of pulverized coal. P1, P2 and P3 are obtained from principal component analysis of the injection characteristic of pulverized coal determined by lab experiment. This model is proved to be effective on both lab experiment coal blending and actual production of BF.