The fibrosis staging criterion (Knodell's) for viral hepatitis in human should not be applied to experimental liver fibrosis in rodents because of differences of species and etiology. Liver fibrosis was induced by carbon tetrachloride (CCl4) or albumin antigen-antibody complex in Balb/C mice or Wistar rats. The degree of fibrosis was quantified by staging scores or hydroxyproline measurement or image analysis. Inter- and intra-observer variations of the criterion were also evaluated. Liver fibrosis was divided into seven stages: stage 0, no fibrosis; stage 1, short fibrous tissue in central venule (C); stage 2, fibrous C-C septa appearance; stage 3, C-C fibrous septa developed incompletely; stage 4, C-C septa completely connected (pseudo-lobule); stage 5, C-P (Portal Tract) bridging fibrosis, nodular appearance <= 50%.; and stage 6, nodular appearance > 50%. There was significant correlation between the staging scores and hydroxyproline concentration (r = 0.708 in mice, r = 0.841 in rats) and, between staging scores and fibrosis area (r = 0.919 in rats, P < 0.001). The interobserver and intraobserver agreement was consistent (kappa = 0.738 or 0.726, P < 0.001). This staging of hepatic fibrosis in rodents is scientifically rational and repeatable, and is therefore suggested as the criterion for the assessment of experimental hepatic fibrosis in rodent studies.