BACKGROUND.The objective of this study was to examine the usefulness of platelet counts in the diagnosis of cirrhosis and for identifying high‐risk individuals in a community‐based hepatocellular carcinoma (HCC) screening program.METHODS.Pilot Study 1 determined the correlation between platelet counts and pathologic hepatic fibrosis scores among individuals with chronic hepatitis B virus (HBV) infection (n = 122 patients) and hepatitis C virus (HCV) infection (n = 244 patients). Pilot Study 2 investigated proportions of individuals with thrombocytopenia (<150 × 103/mm3) among patients with HCC (n = 4042 patients). Pilot Study 3 demonstrated the correlation between platelet counts and ultrasonographic (US) parenchyma scores among anti‐HCV‐positive individuals (n = 75 patients). The core study was a 2‐stage, community‐based screening for HCC among residents age 40 years or older in townships with a high prevalence of anti‐HCV (n = 4616 individuals) and in townships with a low prevalence of anti‐HCV (n = 1694 individuals). Patients with thrombocytopenia were identified for US and α‐fetoprotein screening.RESULTS.Among the individuals who were positive for anti‐HCV, platelet counts decreased according to increased pathologic fibrosis scores or US scores for liver parenchyma disease: The best cutoff platelet count was 150 × 103/mm3 for a diagnosis of cirrhosis. The sensitivity and specificity were 68.2% and 76.4%, respectively, for pathologic cirrhosis and 76.2% and 87.8%, respectively, for US cirrhosis. Forty‐eight percent of patients with HCC were thrombocytopenic. The proportion of thrombocytopenia was significantly greater in patients with HCV‐related HCC (63%) than in patients with HBV‐related HCC (42%). In the townships with high and low anti‐HCV prevalence, the prevalence of thrombocytopenia was 17.9% and 6.1%, respectively, (P < .001), respectively. Twenty‐five patients were diagnosed with HCC, and all of those patients resided in the high‐prevalence township.CONCLUSIONS.Thrombocytopenia was a valid surrogate of cirrhosis and a valid marker for the identification of individuals at high‐risk for HCC, especially in areas that had a high prevalence of HCV. Cancer 2006. © 2006 American Cancer Society.