Pollutant emissions from cars are usually measured on a test bench using driving cycles. However, the use of one unique set of driving cycles to test all cars can be seen as a weak point of emission estimation, as vehicles could conceivably be tested differently depending on their performance levels and usage characteristics. A specific study was then conducted to characterize driving conditions and vehicle usage as a function of vehicle categories, as well as to derive driving cycles specially designed for high- and low-powered cars which have significantly different driving conditions. Pollutant emissions were measured on a sample of 30 passenger cars, using on the one hand the three real-world ARTEMIS driving cycles (urban, rural road and motorway), representative of European driving, and on the other hand specific driving cycles. The comparison of the resulting aggregated emissions demonstrates that the usual test procedure (i.e. with a unique set of driving cycles) can lead to strong differences in emissions, particularly for the most recent vehicle categories.