Bottlenecks have been widely studied for uninterrupted flow. The study of bottlenecks in urban traffic networks implies a challenge, since delays and queuing are implicit in interrupted flows. The objective of this paper is to discuss bottlenecks in urban traffic networks and present a methodology to identify them, based on GPS (Global Position System) data from taxis in Bogotá, Colombia. In networks, where stops are frequent, the principle we adopted deals with finding recurrent low-speed sections, beyond expected delays. Urban bottlenecks occur in those road network segments that perform poorly in terms of speed, compared with upstream and downstream conditions, producing recurrent and larger than normal delays. The GPS devices reported, via GPRS (General Packet Radio Service), information every 10 s. Results of this exploratory project are promising. The method allowed the identification of six urban bottlenecks out of seven randomly selected low-speed sections. The most valuable application of this methodology is the prioritization of resources investment in traffic infrastructure improvements. As a low cost option, this exploratory method might be especially attractive in identifying critical points in traffic networks in developing cities, without expensive traffic-monitoring systems.