Objective: To investigate whether craving and demographic factors to predict relapse in alcohol dependence.Methods: This study was a prospective cohort study. From August 2017 to August 2018, 158 Han male inpatients who met the diagnositic and statistical manual disorders-fourth version(DSM-Ⅳ) alcohol dependence diagnostic criteria were recruited from three mental hospitals in China. The participants were interviewed at baseline and followed up by telephone after 3 months for assessment. The baseline assessment after the acute withdrawal period included demographic data and alcohol-related data, clinical institute withdrawal assessment-advanced revised (CIWA-Ar), withdrawal and cue-induced craving on visual analog scale (VAS), Michigan alcoholism screening test (MAST), self-rating anxiety scale (SAS), self-rating depression scale (SDS) and alcohol urge questionnaire (AUQ). According to the follow-up results, “relapse” was defined as the consumption of beverages containing ethanol at any time during the follow-up study, and “time to relapse” was defined as the number of days from the first drinking to the baseline. Whether relapse occurred and the time to relapse were the primary endpoints. Cox proportional hazard regression model was used to analyze the factors affecting the relapse of alcohol dependence. Results: In the study, 158 alcohol dependence patients were finally included, age from 21 to 60 years, with the mean age of (40.31±9.14) years. The relapse rate was 63.7% three months after baseline assessment. According to Cox univariate analysis and multivariate analysis, the age (OR=0.975, P=0.030) and CIWA-Ar scores (OR=1.126, P=0.010) significantly predicted relapse. And there was no significant difference in education level, marital status, withdrawal and cue-induced craving on VAS, SAS and SDS between the relapse group and the non-relapse group (P>0.05).Conclusion: Age and severity of alcohol-dependent withdrawal symptoms during hospitalization are significantly related to relapse for alcohol in alcohol-dependent patients. To be exact, the older age is a protective factor, that is to say, the younger patients are prone to relapse, while the risk of relapse is raised by the higher se-verity of withdrawal symptoms. However, neither cue-induced nor withdrawal craving can predict relapse of alcohol-dependent patients.