We analyze the conditions for nucleation of a dip-slip fault in intact crust and its subsequent growth using linear elastic fracture mechanics and the finite element method. We assume a fault can be modeled as a mode II shear crack in a layered elastic crust, and we investigate fault behavior under conditions of a variable dip and shear fracture energy. Two shear fracture energy gradients, Gc, were evaluated; both have surface values of 1¿105 J/m2 and increase linearly to values of 1¿106 or 1¿107 J/m2 at 15 km depth, the base of the seismogenic zone. We assume that the vertical and horizontal components of the gravitational stress are both equal to the lithostatic load, effectively removing them from the calculation of shear stress on the fault. Strain required for failure varied with respect to shear fracture energy gradient and rupture direction but not with dip. Rupture upward from the base of the seismogenic zone required more than twice the initial strain to cause fault growth and would result in very large earthquakes. Rupture downward in models associated with the higher Gc gradient resulted in stable growth of the fault; that is, additional strain was needed to allow for continuous fault growth from the surface to the base of the seismogenic zone. We suggest that fault nucleation and growth are a stable process which would preferentially occur by rupturing downward under stable conditions. Rerupturing of an established fault may initiate at depth or the surface depending on the degree of healing between rupture events and the tectonic stress available. A vast majority of shallow crustal earthquakes nucleate at the base of the seismogenic zone and rupture upwards. This suggests that most events are a reactivation of an old established fault and not growth of a new or developing fault. ¿ American Geophysical Union 1992