Several arguments suggest that most hepatocellular carcinomas (HCCs) occurring in human cirrhotic livers arise from large hepatocellular nodules or macronodules. Except for nodules with obvious features of HCC, there exist no consistent criteria enabling the differentiation between benign regenerative and neoplastic, potentially malignant macronodules. Surrogate markers able to accurately discriminate those lesions that will evolve toward a HCC are required. In this study, we investigated the clonality of 26 macronodules isolated from eight cases of explanted cirrhotic livers in women by analyzing X-chromosome inactivation, as indicated by the methylation status of the human androgen receptor gene (HUMARA). For each macronodule, a large set of pathological features was evaluated and used to classify the macronodules into four groups: entirely benignlooking nodule (type 1), low-grade dysplastic nodule (type 2), high-grade dysplastic nodule (type 3), and HCC (type 4). Clonal analysis showed that 14 macronodules (54%) were monoclonal and 12 (46%) were polyclonal. Monoclonality was detected in 5 of 11 (45%) nodules from groups of entirely benign-looking and low-grade dysplastic nodules (types 1 and 2) and in 9 of 15 (60%) nodules from the group of high-grade dysplastic nodule and HCC (types 3 and 4). Neither the etiology of cirrhosis nor the size or histological classification of macronodules was correlated with the clonal status. In conclusion, clonal analysis of macronodules enables the differentiation between mono-and polyclonal macronodules in cirrhosis. Because monoclonal macronodules are prone to evolve to HCC, the determination of the clonal status of a macronodule could provide additional information for evaluating the prognosis of these lesions. (HEPATOLOGY 1998;28:953-958.)