Expansion of pore fluid caused by frictional heating might have an important effect on the frictional resistance and temperature during an earthquake and a controlling influence on the physics of the earthquake process. When confined water is heated, the pressure increases rapidly (>10 bars/¿C). As Sibson (1973) has pointed out, this could cause a sharp reduction of effective normal stress and dynamic friction on the fault surface. Whether or not this transient stress reduction occurs depends upon the tandem operation of several processes, any of which can break the chain that links frictional heat to frictional stress; the friction must cause an appreciable temperature rise (imposing conditions on the width of the shear zone and rate of conductive transport); the temperature rise must cause an appreciable fluid pressure rise (imposing conditions on the rate of pore dilatation or hydrofracturing, and the rate of Darcian transport); the fluid pressure rise must cause an appreciable reduction of friction (requiring the presence of a continuous fluid phase). Each process depends upon event duration, particle velocity, and the initial values of dynamic friction. With the present uncertainty in the controlling parameters (principally permeability, width of the shear zone, initial stress, and factors controlling transient hydrofracture and pore dilatation) a wide variety of fault behavior is possible. Limits to fault behavior for various ranges of the controlling parameters can be estimated from the governing equations, however, and results can be summarized graphically. If the effective stress law applies and pore dilataion is unimportant, dynamic friction would drop from an initial value of 1 kbar to ~100 bars when shear strain reached 10 for most earthquakes if the permeability were less than 0.1 μdarcy; the maximum temperature rise would be only ~150 ¿C irrespective of final strain. If the permeability were >100 mdarcies, however, friction would be unaffected by faulting and temperatures could approach melting for shear strains ~20. For permeabilities 1 mdarcy, friction could be reduced appreciably during large earthquakes, but during small ones it could not. Combined with thermal effects, dilatational strain of a few percent of pore volume could lead to virtually frictionless faulting or increasing frictional resistance, depending upon its sign; unstable propagation of hydrofractures (after fluid pressure exceeded the least principal stress) could cause a sudden increase in fault friction. Strengthening due to cooling and Darcian flow at the conclusion of an earthquake could occur in seconds or weeks depending upon event duration, transport parameters, and shear zone width; it might influence the redistribution of stress by aftershocks.