AIMS: A sound predictive test is lacking for the identification of cirrhotic patients at high risk of developing hepatocellular carcinoma. The present study evaluates the measurement of hepatocyte expression of silver stained nucleolar organiser region (AgNOR) proteins as a risk factor for the development of hepatocellular carcinoma in cirrhosis. METHODS: Liver biopsies from 176 cirrhotic patients included in a follow up surveillance programme for hepatocellular carcinoma development were evaluated prospectively for hepatocyte AgNOR protein quantity. The follow up programme consisted of clinical and biochemical assessment every three months, and ultrasound scanning and serum alpha-fetoprotein (alpha FP) assessment every six months. Histological sections from the needle biopsies performed at enrollment were stained selectively for AgNOR proteins and the percentage of hepatocytes with an AgNOR protein area > or = 7 micron 2, indicative of a proliferative state (AgNOR proliferation index (AgNOR-PI)), was measured. RESULTS: During the mean (SD) follow up time of 65.5 (36.29) months (range, 12-143; median, 67), hepatocellular carcinoma was diagnosed in 48 of 176 patients (27.3%). The AgNOR-PI of the whole series ranged from 0% to 5% (median, 0.9%), and was significantly higher in patients with liver cell dysplasia and hepatitis B surface antigen (HBsAg) positivity (p < 0.0001 and p = 0.0002, respectively). The 176 patients were divided into two groups according to their AgNOR-PI scores; a cut off value of 2.5% defined by the receiver operating characteristic curve and the Youden index was used. Forty two patients were included in the high AgNOR-PI (< 2.5%) group, and 134 patients the low AgNOR-PI (< 2.5%) group. In the high AgNOR-PI group, 25 of 42 patients developed hepatocellular carcinoma, in contrast to only 23 of 134 patients (17.2%) in the group with a low AgNOR-PI (p < 0.0001). Hepatocellular carcinoma development was also significantly more frequent in patients with liver cell dysplasia and HBsAg positivity. Multivariate analysis using AgNOR-PI, liver cell dysplasia, HBsAg positivity, and hepatitis C virus (HCV) infection as covariates demonstrated that the AgNOR-PI parameter was the only significant predictor of hepatocellular carcinoma development. CONCLUSIONS: These results demonstrate that a high hepatocyte proliferation rate is a major risk factor for hepatocellular carcinoma development in the cirrhotic liver. Therefore, the evaluation of the hepatocyte proliferation rate is very important to identify patients requiring a more strict follow up programme for early diagnosis of hepatocellular carcinoma.