Driving style evaluation is an important research topic in the field of intelligent transportation, which is usually recognized and classified qualitatively using time or frequency domain analysis methods and lacksobjective, quantitative evaluation system. A quantitative analysis method for driving style of the driver based on phase space reconstruction is proposed. Firstly, the vehicle test data and local neural network are adopted to establish a personalized driver model. Secondly, the personalized driver model is applied to the speed tracking test in normalizeddriving cycle test to achieve the normalization of driving behavior. Finally, the phase space reconstruction of the normalized driving behavior is conducted, a driving style index is proposed based on the correlation dimensionand used for quantitatively evaluating the aggressiveness of the driving behavior. The driving style index is further applied to the recognition of driving styles. Simulation results verify the effectiveness of the proposed method.