The modelling of a clutch in a power train transmission is a delicate process because of its non-linear behaviour. Two different states have to be taken into account: when the clutch is locked and when the clutch is slipping. Moreover the clutch has often to be controlled automatically in Hybrid Electric Vehicles (HEVs). An Energetic Macroscopic Representation (EMR) of a clutch system has been developed for insertion in a complete model of HEV. Both clutch states are genuinely taken into account in a physical way. In this paper, an inversion-based control of the clutch system is defined from its EMR. This control is used in the control of a conventional vehicle before being implemented on HEVs. Experimental results are provided to validate the clutch model and the inversion-based control.