In recent years, the interaction between human driver and Lane Keeping Assistance System (LKAS) has gradually aroused concern. As a result, the concept of personalized LKAS is being put forward. To achieve the concept, driver lane keeping characteristic (DLKC) indices which could distinguish different driver lane keeping behavior are essential for LKAS to be adaptive to personal DLKC. However, there are few researches on DLKC indices for personalized LKAS. Although there are many researches on modeling driver steering behavior, these researches are insufficient to extract DLKC indices. Firstly, most of researches are for double lane change behavior which is different from driver lane keeping. Secondly, few researches on lane keeping behavior only provided model structure and rarely discussed about modeling details, such as how to select model data, how to define driver desired center line and so on. In this paper, DLKC indices for personalized LKAS were deeply researched. Firstly, DLKC indices were determined based on driver lane keeping behavior analysis and the previous study in our laboratory on subjective evaluation of LKAS. DLKC indices consisted of the following three parts: steering return timing, steering return process, and steering return ending. Secondly, lane keeping experiments were conducted using Prescan/Carsim software on driving simulator. Thirdly, DLKC indices were extracted based on statistical modeling method and transfer function modeling method. With statistical method, the indices indicating steering return timing and steering return ending were extracted. With transfer function method, the necessity of data selection and driver desired center line definition were introduced. And then transfer function model parameters were identified and extracted to indicate driver steering return process. In the end, DLKC indices were validated on driving simulator and the results showed that they were in accordance with driver lane keeping behavior.