Background Guidelines recommend screening for hepatocellular cancer (HCC) with ultrasonography. The performance of ultrasonography varies widely. Computed tomography (CT) is less operator dependent. Aim To compare the performance and cost of twice-a-year ultrasonography to once-a-year triple-phase-contrast CT for HCC screening in veterans. We hypothesised that CT detects smaller HCCs at lower overall cost. Method One hundred and sixty-three subjects with compensated cirrhosis were randomised to biannual ultrasonography or yearly CT. Twice-a-year alpha-feto protein testing was performed in all patients. Contingency table analysis using chi-squared tests was used to determine differences in sensitivity and specificity of screening arms, survival analysis with Kaplan-Meier method to determine cumulative cancer rates. Multivariate logistic regression models were used to examine predictive factors. Results Hepatocellular cancer incidence rate was 6.6% per year. Nine HCCs were detected by ultrasonography and eight by CT. Sensitivity and specificity were 71.4% and 97.5%, respectively, for ultrasonography vs. 66.7% and 94.4%, respectively, for CT. Although 58.8% of screen-detected HCC were early stage (Barcelona Clinic Liver Cancer stage A), only 23.5% received potentially curative treatment despite all treatment options being available. HCC-related and overall mortality were 70.5% and 82.3%, respectively, in patients with screen-detected tumour. Overall costs were less for biannual ultrasonography than annual CT. Conclusions Biannual ultrasonography was marginally more sensitive and less costly for detection of early HCC compared with annual CT. Despite early detection, HCC-related mortality was high. These data support the use of biannual ultrasonography for HCC surveillance in a US patient population (NCT01350167).