Tunnel entrance is a traffic accident-prone location, for its frequent change of driving environment in the short space. Speed change behavior is relative to traffic accidents directly, for drivers’ subjective manipulating vehicles. However, the constraints of the human factor (e.g., driving experience) and road environment (e.g., relative distance from the tunnel entrance, road slope type) often affect the output of speed change behavior. This paper focused on studying the speed changes of tunnel’s entrance differing of driving experience, relative distance from the tunnel entrance and road slope type. Speed change behavior was determined by calculating the difference between minimum speed and maximum speed, which leads to the behavior models divided into three types: continuous deceleration, uniform speed and continuous acceleration. Multimodal logit models were utilized to analyze the effect of three main factors mentioned above to speed change behaviors. Evaluation of multimodal logit models showed that speed change behaviors vary according to different driving experience, relative distance from the tunnel entrance and road slope type. Besides, the multimodal logit model after entering the tunnel entrance the model shows better performance than before entering’s. Identifying speed change behaviors based on different driving experience and road environment is important for future road and tunnel designs. Keywords: Traffic safety, driving behavior, driving speed, driving experience, multimodal logit model

CICTP