Most of the Advanced Driver Assistance System (ADAS) applications are aiming at improving both driving safety and comfort. Understanding human drivers' driving styles that make the systems more human-like or personalized for ADAS is the key to improve the system performance, in particular, the acceptance and adaption of ADAS to human drivers. The research presented in this paper focuses on the classification and identification for personalized driving styles. To motivate and reflect the information of different driving styles at the most extent, two sets, which consist of six kinds of stimuli with stochastic disturbance for the leading vehicles are created on a real-time Driver-In-the-Loop Intelligent Simulation Platform (DILISP) with PanoSim-RT®, dSPACE® and DEWETRON® and field test with both RT3000 family and RT-Range respectively. Three physical quantities, the root mean square of vehicle acceleration, the time-to-start and the time gap of each driver, are extracted from test samples, and their mean and variance are used as clustering samples. Then driving styles are defined and classified into three categories via Particle Swarm Optimization Clustering (PSO-Clustering) algorithm. The identification models are built as a Multi-dimension Gaussian Hidden Markov Process (MGHMP), and key parameters of identification models are optimized in orthogonal test method. The consistency of the classification and identification results under DILISP and field test are compared. Test results show that the stimuli set consisted of 4 kinds of non-periodic apparent transient step signals should be the prior selection for classification and identification. What’s more, driving styles can be classified clearly and identified effectively with the accuracy rate above 95% by using the proposed classification and identification strategy.