Background and aims Many models have been developed to predict liver-related events (LRE) in chronic hepatitis B, few focused on compensated HBV-induced cirrhosis. We aimed to describe the incidence of LRE and to determine independent risk predictors of LRE in compensated HBV-induced cirrhosis patients receiving antiviral therapy using routinely available parameters. Methods Prospective cohorts of treatment-naive adults with compensated HBV-induced cirrhosis were enrolled. Patients were treated with entecavir (ETV) or ETV + thymosin-alpha1 (Thy-alpha 1) or lamivudine (LAM) + adefovir (ADV). Data were collected at baseline and every 6 months. LRE was defined as development of decompensation, HCC or death. Results Totally 937 patients were included, 608 patients treated with ETV, 252 with ETV + Thy-alpha 1, and 77 with LAM + ADV. After a median follow-up of 4.5 years, 88 patients developed LRE including 48 with HCC. The cumulative incidence of LRE at year 1, 3, and 5 was 2.1%, 7.0%, and 12.7%, respectively, and was similar for three treatment groups. All models using variables at month 6 or 12 had better fit than models using baseline values. The best model for prediction of LRE used PLT, GGT, and AFP at month 6 [AUC: 0.762 (0.678-0.814)], for hepatic decompensation-PLT, LSM and GGT at month 12 (AUC: 0.834 (0.675-0.919)), and for HCC-AFP and GGT at month 6 [AUC 0.763 (0.691-0.828)]. All models had negative predictive values of 94.0-98.8%. Conclusion Models using on-treatment variables are more accurate than models using baseline variables in predicting LRE in patient with compensated HBV-induced cirrhosis receiving antiviral therapy. ClincialTrials.gov number NCT01943617, NCT01720238, NCT03366571, NCT02849132.