On the validity of the "hard evaporation" approach