Cirrhosis is an asymptomatic liver disease that is often discovered when the patient's condition is irreversible. Therefore, the treatment consists of a series of measures to control the progression of the disease, since the main consequence of the cirrhosis is the increase of the portal venous pressure, which causes the appearance of varices and their respective rupture may be fatal. Studies related to that disease are very important, so the statistical analysis is a tool that helps to make decisions in medical procedures and patient follow-up. The most applied statistical method in the biomedical sciences is survival analysis, which consists of describing the time of occurrence until the event of interest. However, there are situations in which a proportion of the sample does not experience the interest outcome, even if they are accompanied by a long period of time. In such cases, such observations are said to be immune to the outcome of interest and traditional survival analysis methodologies are not appropriated. Therefore, the care fraction or long duration models can be used in these situations because they incorporate mixtures of models to solve the complexity inherent in the actual study. In this work, some statistical models were considered to analyze the survival times of patients, after surgery of Endoscopic Band Ligation of the Esophageal Varices, such as Exponential, Gamma and Weibull models. The data is related to survival times of 129 patients, who were treated in the Hospital das Clínicas of the Medical School of UNESP, Campus Botucatu (SP), from 2006 to 2010. The Modi ed Weibull distribution with cure rate was considered adequate to the data, indicating that older male patients with higher Child-Pugh score without taking beta blockers medicine are more likely not to be a long-term survivor.