Hepatocellular carcinoma (HCC) surveillance is underutilized among patients with cirrhosis. Understanding which steps in the surveillance process are not being conducted is essential for designing effective interventions to improve surveillance rates. The aim of our study was to characterize reasons for failure in the HCC surveillance process among a cohort of cirrhotic patients with HCC. We conducted a retrospective cohort study of cirrhotic patients diagnosed with HCC at a large urban safety-net hospital between 2005 and 2011. Patients were characterized by receipt of HCC surveillance over a two-year period before HCC diagnosis. Among patients without HCC surveillance, we classified reasons for failure into four categories: failure to recognize liver disease, failure to recognize cirrhosis, failure to order surveillance, and failure to complete surveillance despite orders. Univariate and multivariate analyses were conducted to identify predictors of failures. We identified 178 patients with HCC, of whom 20% had undergone surveillance. There were multiple points of failure-20% had unrecognized liver disease, 19% had unrecognized cirrhosis, 38% lacked surveillance orders, and 3% failed to complete surveillance despite orders. Surveillance was more likely among patients seen by hepatologists [OR, 6.11; 95% confidence interval (CI), 2.5-14.8] and less likely in those with alcohol abuse (OR, 0.14; 95% CI, 0.03-0.65). Although a retrospective analysis in a safety-net hospital, our data suggest that only one in five patients received surveillance before HCC diagnosis. There are multiple points of failure in the surveillance process, with the most common being failure to order surveillance in patients with known cirrhosis. Future interventions must target multiple failure points in the surveillance process to be highly effective.