Fibrosis is a common endpoint of clinical trials in chronic hepatitis C, and liver biopsy remains the gold standard for fibrosis evaluation. However, variability in the distribution of fibrosis within the liver is a potential limitation. Our aim was to assess the heterogeneity of liver fibrosis and its influence on the accuracy of assessment of fibrosis with liver biopsy. Surgical samples of livers from patients with chronic hepatitis C were studied. Measurement of fibrosis was performed on the whole section by using both image analysis and METAVIR score (reference value). From the digitized image of the whole section, virtual biopsy specimens of increasing length were produced. Fibrosis was assessed independently on each individual virtual biopsy specimen. Results were compared with the reference value according to the length of the biopsy specimen. By using image analysis, the coefficient of variation of fibrosis measurement with 15-mm long biopsy specimens was 55%; and for biopsy specimens of 25-mm length it was 45%. By using the METAVIR scoring system, 65% of biopsies 15 mm in length were categorized correctly according to the reference value. This increased to 75% for a 25-mm liver biopsy specimen without any substantial benefit for longer biopsy specimens. Sampling variability of fibrosis is a significant limitation in the assessment of fibrosis with liver biopsy. In conclusion, this study suggests that a length of at least 25 mm is necessary to evaluate fibrosis accurately with a semiquantitative score. Sampling variability becomes a major limitation when using more accurate methods such as automated image analysis.