A simple cyclic process is proposed to explain why major strike-slip fault zones, including the San Andreas, are weak. Field and laboratory studies suggest that the fluid within fault zones is often mostly sealed from that in the surrounding country rock. Ductile creep driven by the difference between fluid pressure and lithostatic pressure within a fault zone leads to compaction that increases fluid pressure. The increased fluid pressure allows frictional failure in earthquakes at shear tractions far below those required when fluid pressure is hydrostatic. The frictional slip associated with earthquakes creates porosity in the fault zone. The cycle adjusts so that no net porosity is created (if the fault zone remains constant width). The fluid pressure within the fault zone reaches long-term dynamic equilibrium with the (hydrostatic) pressure in the country rock. One-dimensional models of this process lead to repeatable and predictable earthquake cycles. However, even modest complexity, such as two parallel fault splays with different pressure histories, will lead to complicated earthquake cycles. Two-dimensional calculations allowed computation of stress and fluid pressure as a function of depth but had complicated behavior with the unacceptable feature that numerical nodes failed one at a time rather than in large earthquakes. A possible way to remove this unphysical feature from the models would be to include a failure law in which the coefficient of friction increases at first with frictional slip, stabilizing the fault, and then decreases with further slip, destabilizing it.