Summary: In clinical medicine and epidemiology, the mean life of patients after a certain treatment is of interest. Due to the censor, the estimation of survival function tail is not accurate. Alternatively, the restricted mean life is usually considered. For measuring the difference of treatment, under non-randomized grouping, we estimate the restricted mean life while the independent and dependent censor data exist simultaneously. We construct the additive hazard model for explaining dependent variate and proportional hazard model for survival time. Inverse probability censoring weighted method is used to estimate the parameters of the models and the large sample of the estimator is discussed. The numerical simulation of parameter estimation is presented. Finally, the method we provide is applied to liver transplant data.