De-capacity in heavy industry is the key and difficult point of China’s supply side reform.Due to the lack of traceable historical and structural data on capacity utilization by each industry,the existing studies have many differences in the historical causes,structural evolution and main crux about the overcapacity.Based on the Translog Cost Function and with the estimation on the seemingly uncorrelated regression(SUR),this study uses the microcosmic enterprise data supplied by the China Industry Enterprise Database to try to provide the unified data base which can effectively characterize the historical evolution and structural characteristic of capacity utilization in different industries,and selects three representative sectors(steel,coal and cement)for empirical comparison.The research result shows that(1)the turning point in the capacity pattern of China’s heavy industry roughly appeared in 2004,and the subsequent expansion in supply and policy interference led to the gradual emergence of overcapacity,which had the most impact on the steel and other fields with high withdrawal cost;(2)The model of supply side reform should be different:De-capacity in steel industry needs more administrative intervention,the key de-capacity of coal industry is supply optimization,and the capacity of cement industry can be automatically adjusted by market mechanism;(3)The market-oriented reform of state-owned enterprises is not the only good medicine for de-capacity,the industrial integration based on the industry characteristics will improve overall risk resistance.The study can also be used to monitor and analyze the overcapacity in other industries.