Abstract: Precision medicine emphasizes the importance of correctly identifying heterogeneous subgroups to develop individualized treatment strategies that can be prescribed for each subgroup. Despite the recent methodological advances in subgroup analysis, how to effectively identify subgroups for censored data remains largely unexplored. In this paper, we propose a new methodology for subgroup analysis based on the accelerated failure time model, in which the heterogeneity caused by latent factors can be represented by subject-specific intercepts. We consider the most common right censoring situation, and processes the censored data by the mean imputation method. The hard thresholding penalty function is applied to pairwise differences of the intercepts, thus automatically dividing the observations into different subgroups. We also establish the theoretical properties of our proposed estimator. Our proposed method is further illustrated by simulation studies and analysis of a wisconsin breast cancer dataset.