Background: Sorafenib is a promising drug for advanced hepatocellular carcinoma (HCC); however, treatment may be discontinued for multiple reasons, such as progressive disease, adverse events, or the cost of treatment. The consequences of sorafenib discontinuation and continuation are uncertain. Materials and Methods: We retrospectively analyzed 88 HCC patients treated with sorafenib from July 2007 to January 2013. Overall survival (OS), post-disease progression overall survival (pOS), and time to disease progression (TTP) were compared for survival analysis. Cox proportional hazard regression was performed to assess the effect of important factors on OS in the overall patient population and on pOS in patients who continued sorafenib treatment. Results: Sorafenib was discontinued and continued in 24 and 64 patients, respectively. The median OS (355 vs 517 days respectively; p=0.015) and median post-PD OS (260 vs 317 days, respectively; p=0.020) were statistically different between the discontinuation and continuation groups. Neither the median time to first PD nor the time to second PD were significantly different between the 2 groups. In the discontinuation group, 3 of the 24 patients (12.5%) suffered disease outbreaks. In Cox proportional hazard regression analysis after correction for confounding factors, BCLC stage (p=0.002) and PD site (p=0.024) were significantly correlated with pOS in patients who continued sorafenib treatment. Conclusions: Sorafenib discontinuation may cause HCC flares or outbreaks. It is advisable to continue sorafenib treatment after first PD, particularly in patients with Barcelona Clinic Liver Cancer stage B disease or only intrahepatic PD.