Summary form only given, as follows. Urban traffic numeric simulations are currently performed on the base of a relatively little knowledge, since field vehicle data collection is expensive and limited in both space and time. On the other hand, aerial stereoscopic photography allows detection of vehicles simultaneously and over large areas. However, extracting vehicle information from pictures in an urban context, while filtering-out undesirable features at the ground, is a difficult task: in this work we propose and implement an innovative processing method for vehicle identification and counting. So aerial pictures could be used to effectively improve mathematical traffic model output. The methodology can be easily applied over the whole road-network at a relatively low cost and fast data collection time. © 2001 IEEE.