Background: Data are insufficient regarding the survival benefit of surveillance for hepatocellular carcinoma (HCC). Aim: To investigate the effectiveness of HCC surveillance in a hepatitis B-endemic population. Methods: This retrospective cohort study included 1402 consecutive patients who were newly diagnosed with HCC between 2005 and 2012 at a single tertiary hospital in Korea. The primary endpoint was overall survival. Lead-time and length-time biases were adjusted (sojourn time = 140 days) and sensitivity analyses were performed. Results: The most common aetiology was hepatitis B (80.4%). Cirrhosis was present in 78.2%. HCC was diagnosed during regular surveillance (defined as mean interval of ultrasonography <8 months, n = 834), irregular surveillance (n = 104) or nonsurveillance (n = 464). Patients in the regular surveillance group were diagnosed at earlier stages ([very] early stage, 64.4%) than the irregular surveillance (40.4%) or nonsurveillance (26.9%) groups and had more chance for curative treatments (52.4%) than the irregular surveillance (39.4%) or nonsurveillance (23.3%) groups (all P < 0.001). Mortality risk was significantly lower in the regular surveillance group (adjusted hazard ratio [aHR], 0.69; 95% [CI], 0.57-0.83) but not in the irregular surveillance group (aHR, 0.94; 95% CI, 0.69-1.28) compared with the nonsurveillance group after adjusting for confounding factors and lead-time. When the subjects were restricted to cirrhotic patients or Child-Pugh class A/B patients, similar results were obtained for mortality risk reduction between groups. Conclusions: HCC surveillance was associated with longer survival owing to earlier diagnosis and curative treatment. Survival advantage was significant with regular surveillance but not with irregular surveillance.