The Real Driving Emission (RDE) procedure will measure the pollutants, such as NOx, emitted by cars while driven on the road. RDE will not replace laboratory tests, such as the current WLTP but it will be added to them. RDE is complementary to the laboratory-based procedure to check the pollutant emissions level of a light-duty vehicle in real driving conditions. This means that the car will be driven on a real road according to random acceleration and deceleration patterns conditioned by traffic flow. So, the procedure will ensure that cars deliver real emissions over on-road and so the currently observed differences between emissions measured in the laboratory and those measured on road under real-world conditions, will be reduced. However, the identification of a path on the road to check the test conditions of RDE is not easy and hardly repeatable. In fact, when the car is driven by a real driver on a real road we need, firstly, to overcome some problems for identifying the best way to travel through an unknown traffic area to respect the "conformity factor" in the RDE procedure. In this paper, the pollutant emissions of three different Euro 5 cars are measured by portable emission-measuring systems (PEMS) complying with RDE specification. The PEMS configuration and other experimental acquisition devices are similar for all three cars. In addition, the statistical analysis of kinematic data is presented to evaluate road gradient variation. The experimental plan includes some repetitions of the urban section in Naples city center and a ring road path. In particular, we also try to obtain and to evaluate the gradient effect (positive and negative) on representative urban cycle (flat) and on a rural one. Moreover, a statistical analysis of emission factors is performed on the urban and rural section.