Liver biopsy is used as the "gold standard" for the assessment of the stage and degree of activity in chronic hepatitis C and is of major importance in evaluating the effects of treatment. Because numerous therapeutic trials are undertaken with histological control, the reproducibility of liver biopsy interpretation appears essential. Therefore the aim of this study was to estimate intraobserver and interobserver variations in the assessment of features, classification, and numerical scoring of chronic viral hepatitis C among 10 pathologists specializing in liver diseases. These pathologists independently reviewed 30 liver biopsy specimens of viral hepatitis C and completed a histological form for each of the specimens. Five pairs of pathologists then were randomly designated. They independently reviewed the biopsy specimens and filled out a new coding form. The interobserver variation was calculated for each item among the 10 individuals and then among the five pairs with the intraclass correlation coefficient or K statistics. Five features showed an almost perfect or a substantial degree of concordance among the 10 observers (Le., cirrhosis, fibrosis, fibrosis grading of Knodell index, steatosis, portal lymphoid aggregates). The 17 other indicators showed a weaker concordance with, for instance, a moderate degree of concordance for piecemeal necrosis, disease activity, Knodell index, a fair degree of concordance for lobular necrosis, and only a slight degree of concordance for six items. Five items had a higher concordance when viewed by a pair of pathologists than when studied by only one pathologist (i.e., steatosis, periportal necrosis grading of Knodell index, lobular necrosis grading of Knodell index, centrilobular fibrosis, and ductular proliferation). Two showed a significant decrease, and the others were unchanged. This study reveals that certain features of mqior importance in assessing disease