Sorafenib improves overall survival (OS) of patients with hepatocellular carcinoma (HCC) in the absence of objective response. Thus, time to tumor progression (TTP) is used to capture benefits of novel molecular agents, but proof of its surrogacy with survival is lacking. Furthermore, survival predictors upon progression are not established and there is a need to characterize postprogression survival (PPS) and assess with time-dependent covariates analysis if it is influenced by progression pattern, and not solely by simultaneous impairment of liver function and performance status. We prospectively followed HCC patients treated with sorafenib. Clinical and biochemical evaluation were done every 4 weeks. Radiologic assessment of progression was done at week 4 and then every 8 weeks using RECIST 1.1. The progression pattern was divided into: intrahepatic/extrahepatic increase in tumor size, new intrahepatic lesion, and new extrahepatic lesion (NEH). We included 147 patients (hepatitis C virus [HCV] 57.1%, performance status [PS] 0 83.6%, Child-Pugh A 82.3%, and BCLC-C 47.3%). The median OS was 12.7 months and its independent predictors (hazard ratio [HR], 95% confidence interval [CI]) were: baseline BCLC 2.49 [1.66-3.73], PS 1.86 [1.12-3.10], registration during follow-up of Child-Pugh B or Child-Pugh C scores (2.36 [1.51-3.69] and 2.89 [1.62-5.15], respectively), definitive sorafenib interruption 2.48 [1.54-4.01], and TTP 3.39 [1.89-6.1]. The presence of NEH 2.42 [1.32-4.44] is also an independent predictor of OS and PPS in patients with radiologic progression. Conclusion: Tumor progression is a surrogate of survival but its impact varies according to progression pattern. Thus, PPS is influenced by progression pattern and this is key in prognostic prediction and second-line trial design and analysis. © 2013 by the American Association for the Study of Liver Diseases.