Earthquake dynamics, plate motions, topographic gradients, and frictional heating are all influenced by the level of shear stresses resolved on crustal faults. However, because few in situ stress measurements are made at depths beyond a few kilometers below the surface, the magnitude of these stresses and the ultimate strength of seismogenic faults remain poorly understood. Here I use mechanical models to show that earthquake rupture properties such as the static stress drop and mean coseismic slip, scale predictably with the stress state on faults and the friction drop during rupture. These models show that if faults have stress states defined by Byerlee-level friction and hydrostatic fluid pressures, then static stress drops and average coseismic slip amplitudes should increase with down-dip rupture width, and be consistently greater for reverse ruptures than for normal fault ruptures of the same dimension (by a factor of ∼3). Although sometimes observed, these trends are inconsistent with most earthquake data. One possible explanation for this discrepancy is that most faults in the crust experience anomalously low, and nearly depth-independent, shear stress states, which could be due to fluid overpressures or intrinsically weak fault rocks at depth. However, it is difficult to rule out the possibility that the model predictions cannot currently be resolved from earthquake data obtained from seismic methods.