Synopsis: Due to low cost, clean working environments, linear power transfer facilities and good performance, pneumatic systems play a significant role in industrial processes. During the last twenty years, the parallel development of pneumatic systems and control theory has lead to the implementation of modern control laws in pneumatic devices [Edge 1997]. The aim of this paper is to present and to compare experimental results for an electropneumatic positioning system. A classical linear control law with scheduling gains, established with the tangent linearized model, is compared with a more sophisticated nonlinear control law based on input-output linearization [Isidori 1989]. The tests are realised with two different admissible trajectories. The first one is a trajectory with a constant stage, which is usually used in robotic, to control painting system for example. The second one is a trajectory with constant acceleration and deceleration stages, often used for positioning systems which carry delicate loads or dangerous liquids in containers (as in the chemical industry). This choice of the trajectories has also to be related to a problem of sizing. The experimental results are presented in terms of repeatability for each control law implemented on the same device: an in line electropneumatic servo-drive. The statistical comparisons between nonlinear and linear control laws are shown in terms of static error: mean value, standard deviation value, minimum and maximum values. The precision and the repeatability are deduced from these results. All these indicators are also processed during the dynamical stage. The customer for choosing an appropriate control law for its desired specifications may use all these results.