Abstract: The caustic technique uses galaxy redshifts alone to measure the escape velocity profiles of galaxy clusters to clustrocentric distances beyond their virial radius, where dynamical equilibrium does not necessarily hold. This feature allows us to estimate the mass profile and to identify the members of a cluster. We analyze the ability of this technique to extract these dynamical properties and we study its possible systematic errors, by applying it to simulated clusters with mass M200 larger than 10^14 Msun/h extracted from a cosmological hydrodynamical simulation of a LCDM universe. We evaluate the systematic effects introduced by the center identification, the choice of the parameters and the projection effects and we analyze the dependence of the profiles on the number of galaxies present in the catalogs.