Traffic congestion is the most common phenomenon observed on the road network.It is also the simplest indicator to describe the traffic conditions. This study explores the subjectively perceived traffic congestion of the drivers on arterials. To analyze the cause effects of driver’s behavior upon the time and space dimensions, we apply one order SEM (structural equation model) plus Catastrophe theory. To make the outcomes reliable, some well setup methodology of habit questionnaire design is adopted. Not only can the quality of currently used congestion be significantly improved, but also the best locations of congestion information are found. Finally, to test the model transferability, two areas of route 15 and arterials in Taipei city are selected, followed by promising performance evaluation discussed in details.