Road transport is the main contributor of emissions through vehicle exhaust. Vehicle exhaust emissions exclusively depend on vehicle characteristics, road geometry, traffic characteristics, atmospheric conditions, vehicles’ age and driving behaviour. In the road network, intersections are crucial elements because of their regulation, traffic composition and geometric features. Generally, the driver has to decrease speed and sometimes stop for a long period of time while approaching the intersections, which cause a high concentration of emissions. It is a prior need to study the speed pattern of vehicles at intersections in the wake of fundamental for air quality analysis. To comprehend the driving characteristics of an individual vehicle and speed variation at the intersection, the driving cycle is an important concept being used from many years. Driving cycle is the speed-time profile of the vehicle in stipulated traffic condition. To represent the driving characteristics of an area or a city, the development of the driving cycle is an appropriate approach. It is the process of generating candidate driving cycles associated with vital driving parameters such as acceleration, deceleration, cruise and idle. Numbers of methods have been proposed for the development of the candidate driving cycles based on the purpose of the analysis. In the present study, micro-trip based driving cycle analysis approach is used for urban intersections of Vadodara city, India. Micro-trips are defined as speed profile of the vehicles from upstream to downstream of the intersection, it is the stretch at which succession driving states deceleration, idle and acceleration observed distinctly in the vicinity of intersections. To collect the speed data, the V-box (Velocity box) is embedded on the test vehicles of motorized three wheelers, motorcycle and car. The result shows that the percentage of time spent in acceleration-deceleration state for motorized three wheelers relatively 45% greater than motorcycle and car as 25%. The idling period is high for motorcycle and car ranges from 42-45%. Validation of candidate driving cycles is accomplished through the sum squared difference between data set.