Fuel consumption and pollutant emissions from vehicles are mainly a function of the vehicle's use (journey type. frequency, etc...) and of the vehicle's operating conditions (speed, engine speed, rates of acceleration, temperature conditions, etc...) and depend on both the traffic conditions and the individual behaviour of the driver. Thus, a realistic assessment of emissions, pollution reduction methods and the effectiveness of emission control technologies cannot be carried out without taking into account the actual operating conditions of the vehicles. For that reason, a project modem (Modelling of Emissions and Fuel Consumption in Urban Areas - DRIVE project UI053) has been carried out within the EC DRIVE (Dedicated Road Infrastructure for Vehicle Safety in Europe - EEC research programme - DGXIII) programme to develop models of vehicle emissions in terms of traffic operations. Part of the project was a large scale experiment to measure engine and vehicle operating conditions during normal driving. A total of 58 private cars were equipped with sensors and data acquisition systems to record details of their use by their owners over a period of about 1 month. Over 8200 trips were monitored, covering more than 73 000 km of travel, during which the vehicle speed, engine speed, temperatures and other variables were recorded at 1-s intervals. 'Kinematic sequences' were defined as the periods between two successive stops of the vehicle and these were used to characterize the traffic conditions, which were assumed to be homogeneous throughout each 'sequence'. The main categories of sequences (urban, non-urban and motorway) were identified by factorial analysis and classifying techniques and each sequence was assigned to a category. An analysis of the way in which the sequences were linked during a trip and statistics on trip lengths were used to develop realistic and statistically significant test cycles representing different urban traffic conditions. The cycles were created by recombining real sequences, randomly selected in accordance with the results of the statistical analyses. These cycles were then used on a chassis dynamometer for accurate measurements of exhaust emissions and fuel consumption on a large sample of vehicles. This paper describes the statistical analysis of the vehicle operation data and the development of the test cycles.