Hemophilia B is a hereditary disease caused by defects in coagulation factor IX. Accumulation of sequence data in the hemophilia B database makes it possible to study this disease at the molecular level. The most common mutations reported in the database are amino acid substitutions. Activity of factor IX in a patient's blood depends on a position of the substitution and combination of original and substituting amino acids. We have developed a method to estimate the effects of amino acid substitutions in factor IX of hemophilia B patients. We adopted two kinds of scores: (1) amino acid homology matrices, and (2) distances obtained from physical-chemical properties of amino acids. We found that: (1) the distance obtained from molecular volume shows the highest correlation with factor IX activity among four physical-chemical properties of amino acids, and (2) Dayhoff's 250PAM matrix gives the highest correlation among three homology matrices. We performed a multiple regression analysis for the estimation of factor IX activity by using four amino acid parameters and obtained a distance matrix, which gives the multiple correlation coefficient 0.6. Hemophilia is a hereditary, X-linked, recessive hemorrhagic disorder. About three-fourths of patients with hemophilia have the A type which is due to deficient factor VIII activity. The B type is less frequent than the A type and is due to deficient factor IX activity. Factor IX is a vitamin K dependent plasma protein that participates in the middle phase of blood coagulation. The entire nucleotide sequence of the factor IX gene was reported by Yoshitake et. al. There have been reported a variety of defects in the factor IX gene from hemophilia B patients, and these are summarized in the hemophilia B database. The effects of amino acid substitutions on the activities of factor IX depends on two factors. One is the combination of native and hemophilic amino acids, the differences of physical-chemical properties of the amino acids. Another is the position of mutation in factor IX protein. In this study, we analyzed missense mutations in the database described with factor IX activity values. Among them, the cases with double mutations were excluded from the analysis, leaving 255 cases. The number of unique mutations are 127. We have introduced distances between 20 amino acids by using the following four physical-chemical properties: (1) Molecular volume, (2)Hydropathy, (3) Polar requirement, and (4) Isoelectric point. We also adopted three homology matrices. These matrices are symmetric and composed of 20 x 20 elements, corresponding to amino acid pairs: (1)Dayhoff's 250 PAM matrix, (2) SCM (Structure-derived Correlation Matrix), and (3) SSM (Structure Superposition Matrix). This analysis shows that the highest correlation with factor á¿ activity is 0.47 given by molecular volume. The second one is 0.4 of Dayhoff's 250 PAM matrix. The multiple regression analysis, using four physicalchemical properties of amino acids, gives a correlation coefficient of 0.60. The amino acid distance obtained by the multiple regression analysis provides a useful measure to estimate the severity of the disease caused by a missense mutation. For example, if the distance between native and hemophilic amino acids is less than 1.5, the ratios of severe cases (activity is less than or equal to 1% of the normal), moderate cases (less than or equal to 4%), and mild cases are 67%, 27% and 6%, respectively.