Abstract: In order to explore the efficiency of drivers’ visual recognition of traffic signs, an evaluation system of visual recognition efficiency of road signs in scenic spots was constructed. The field eye-tracking experiment was designed and data of drivers’ visual recognition of traffic signs in scenic spots were collected. Based on the data variation rule and index characteristics, the optimal value and the critical value of the eye movement characteristic data were analyzed and the evaluation index set of traffic sign recognition efficiency was established. Then, support vector machine-recursive feature elimination was used to rank the significance of evaluation indexes to determine the degree of influence of each evaluation index on the evaluation system. In addition, the calibration index of efficacy coefficient method was adopted to establish the evaluation system of visual recognition efficiency. Finally, the system was used to evaluate the visual recognition in Nanjing Zhongshan Mountain National Park and some suggestions on improvement and optimization of traffic sign setting from the perspective of visual recognition efficiency were proposed.