AIMS: To evaluate sampling variability of liver biopsy in patients with primary biliary cirrhosis (PBC). METHODS: Sections from 50 PBC liver specimens obtained at transplantation were examined. The degree of fibrosis was assessed on a scale of 0-4 using two methods: (1) simulated needle biopsy in fields approximately the size of conventional needle biopsy; and (2) whole section scanning in areas with little and extensive fibrosis. RESULTS: Considerable variation in the range of stages of fibrosis was found when the whole section scanning method was used. Only 10 (20%) samples had a consistent degree of fibrosis in all sections scanned. By contrast, the same fibrosis stage was assigned in 30 (60%) specimens examined using the simulated needle biopsy method. When the results obtained by the two methods were compared, there was a discrepancy of one or two stages in 32 samples. This discrepancy was the result of discovering areas with a lesser degree of fibrosis in whole sections compared with the simulated needle biopsy specimens. CONCLUSION: There is considerable variation in the degree of fibrosis in the livers of patients with PBC, even when end stage specimens obtained at transplantation are examined. Consistent results were obtained when simulated needle biopsy specimens were examined. This, however, may be a reflection of the procedure applied when staging liver needle biopsy specimens, where the greatest degree of abnormality is used in determining the stage. The practice of staging of PBC in small needle biopsy specimens is valuable as long as the appearances are interpreted with caution, bearing in mind that there is considerable variability in the degree of fibrosis.