Abstract: In order to explore the statistics correlation between blending coal and mechanical strength of coke, multiple linear regression analysis is made using SPSS software with the coal quality parameters of Ash (A_d), volatile matter (V_daf), Total sulfur (S_{t, d}) and caking index (G) as independent variables and coke′s crushing strength (M₄₀), wear-resistant strength (M₁₀), reactivity index CRI and post-reaction strength CSR as dependent variables. Based on result of analysis, a prediction model is developed. As evidenced by test result, the use of the method can produce a quick and accurate result, and is suitable for making coke quality prediction and control.