A point in the state space of a stochastic process is called “thin” if the associated occupation measure of the ball centered at this point is exceptionally small. We consider in this talk the Hausdorff dimension of thin points of a class of Markov processes with jumps. This is a joint work with Stéphane Seuret.