Thrombocytopenia has been reported as a valid surrogate for liver cirrhosis and could be used to identify groups at high risk of hepatocellular carcinoma (HCC) for ultrasonographic (US) screening. We designed this two-stage community-based screening for HCC. In 2004, subjects (ages z40 years) were invited to undergo comprehensive health examinations, with 17,551 men (ages 63.0 F 11.5 years) and 39,151 women (ages 59.9 F 11.7 years) participating. Subjects with platelet counts <150 Â 10 9 /L or A-fetoprotein (AFP) >20 ng/mL were enrolled for the second-stage US screening; 3,242 subjects (5.7%; male/ female, 1,415/1,827; age 66 F 10 years) were candidates for US screening and 2,983 (92.2%) responded. Of 137 suspected cases, 124 (90.5%) complied with referral for confirmation and 72 (58.1%) were confirmed to be HCC cases (male/female, 41/31; age 68.1 F 8.8 years). Screening with AFP, thrombocytopenia, or both could identify 0.64% (n = 364), 5.33% (n = 3,205), and 5.7% (n = 3,242) of the high-risk subjects from the population, estimated to include 50.5%, 54.5%, and 71.3% of all HCC cases. Among confirmed patients, tumor diameters were <3 cm for the 27 (37.5%) patients and 3 to 5 cm for the 23 (31.9%) patients. Only 5 (6.9%) patients' conditions were too advanced to be actively treated. This study enrolled only 5.7% of the participants for US, which cover 64.7% to 71.3% of the HCC cases. Most (93%) of the detected cases were caught early enough to undergo effective treatment modalities. This HCC screening protocol should be feasible, economical, and effective. (Cancer Epidemiol Biomarkers Prev 2008;17(7):1813 -21)