To investigate whether serum or plasma should be used for the measurement of blood hepatocyte growth factor, the levels were compared in 28 normal subjects and 30 patients who had undergone surgery. The serum level was significantly higher than the plasma level. The serum and plasma hepatocyte growth factor levels differed markedly depending on the subjects, although overall there was a significant correlation between levels (r=0.862, P=0.0001). In serum obtained by the clotting of platelet- or leukocyte-containing plasma with thrombin, hepatocyte growth factor increased in proportion to the number of leukocytes. The difference between serum and plasma hepatocyte growth factor levels also correlated with the number of leukocytes in the patients (r=0.642, P=0.0004). Such a correlation was not observed for platelets. These findings suggest that the serum hepatocyte growth factor level does not strictly reflect the in vivo blood level, due to the release from leukocytes during sample preparation (i.e., blood clotting ) and that plasma is more suitable for assay of blood hepatocyte growth factor.