On the basis of analysis of tens of thousands of primary signals corresponding to original positions of sample systematically, the degree of porosity, the degree of homogeneity and inclusions of billets can be quantitatively determined. The degree of porosity and the maximum degree of segregation for different billets were calculated accurately in the present paper. The weight ratio of carbon content at each original position of billet was suggested to express the degree of homogeneity of billets. Correspondingly, two models-the total weight ratio of carbon contents within the permissive content range (C₀±R) and the confidence extension ratio of median value (K) at 95 % of confidence limit of weight ratio, were provided for practical billet samples analysis. This analysis technique was characterized by the primary, original and statistical nature of measured information, thus the quantitative statistic distribution analysis of chemical elements and their states in larger scale was realized. The quality differences among the billet samples were explained reasonably by above method. Thus, this technique can be used to judge the quality of billet samples.