This paper deals with the problem of controlling electric vehicles charging in a smart grid. We present an event driven model predictive control (MPC) approach, which aims to find a proper trade-off between the needs of minimizing the cost of energy withdrawal and the error in the tracking of a reference aggregated charging power profile. All in respect of drivers' preferences, technical bounds on the control action (in compliance with the IEC 61851 standard) and grid constraints. The proposed control approach allows "flexible" EV users to participate in demand side management programs, which will play a crucial role for stability and efficiency of future smart grids. Simulation results are provided and discussed in details, showing the relevance of our contribution. © 2013 IEEE.