根据我国城市道路交通环境特点，以大学生为研究对象，应用计划行为理论对行人闯红灯行为进行行为决策建模。首先，根据车流量、道路宽度和有无中央安全岛将行人过街环境分为六类；其次，在原计划行为理论模型上增加行为经验新变量，并分别在每类情境下基于计划行为理论进行问卷设计；再次，对广州在校大学生进行问卷调查，总共回收有效问卷376份，有效率为91.75%。对调查数据进行可靠性分析和验证性因素分析，结果表明数据具有良好的信度和效度，各项目的内部一致性良好且可靠性高。最后，基于计划行为理论建立结构方程模型（SEM），对六类情境行人过街行为进行分析，得到大学生闯红灯行为的六个结构路径图。结果表明，大学生对闯红灯的态度和新增加的行为经验与行为意向没有显著相关性，而主观规范和知觉行为控制则与行为意向显著相关，且知觉行为控制是个强大的预测因子，能预测行为意向，还可以直接作用于实际行为。

This paper proposed a decision-making model of college students running on red signal using the theory of planned behavior considering the characteristics of urban road traffic environment in China. Firstly, six scenarios of road crossing are proposed according to traffic flow, road width and central safety island. Secondly, questionnaire on each scenario was designed according to the theory of planned behavior and the behavioral experience variable. More than 400 college students in Guangzhou were involved in the survey, and valid samples are 397, with an effective rate of 91.75%. The data tests show that the survey data has a good performance on reliability, validity and internal consistence. Finally, structural equation modeling (SEM) is used to analyze college students crossing behavior in six scenarios, and six structural path diagrams are obtained. The results show that the attitudes towards crossing on red signals and the new behavioral experience have no significant correlation to behavioral intention, however, subjective norms and perceived behavioral control are significantly related to behavioral intention. The perceptual behavior control is a powerful predictor of behavioral intention, which can not only predict behavioral intention, but also act directly on actual behavior.