The present paper aims at modeling the dynamic width of the virtual lane of the Powered Two-Wheelers (PTWs) and revealing the critical factors that may influence it, such as the speed of the motorcycles, the existence of a heavy vehicle and so on. All possible manners of PTW maneuvering in an urban arterial (2 lanes per direction of travel) were identified and then data were collected from real measurements using advanced video analysis techniques. Further, statistical models were developed in order to reveal the relation between the virtual lane's width and several determinants such as PTW speeds, lane headways, traffic mix and so on. Results showed that the virtual lane width depended on the speed of the PTW, the existence of a platoon, the existence of a heavy vehicle and the headways in both lanes. Moreover, results revealed the need to address the human behavior and the cognitive aspects of driving during the modeling.