A computational study of sliding blocks on inclined surfaces is presen
ted. Assuming that the friction coefficient mu is a function of positi
on, the probability P(lambda) for the block to slide down over a lengt
h lambda is numerically calculated. Our results are consistent with re
cent experimental data suggesting a power-law distribution of events o
ver a wide range of displacements when the chute angle is close to the
critical one, and suggest that the variation of mu along the surface
is responsible for this.