Approximately 100 heat flow measurements in the San Andreas fault zone indicate (1) there is no evidence for local frictional heating of the main fault trace at any latitude over a 1000-km length from Cape Mendocino to San Bernardino, (2) average heat flow is high (~2 HFU, ~80 mW m-2) throughout the 550-km segment of the Coast Ranges that encloses the San Andreas fault zone in central California; this broad anomaly falls off rapidly toward the Great Valley to the east, and over a 200-km distance toward the Mendocino Triple Junction to the northwest. As others have pointed out, a local conductive heat flow anomaly would be detectable unless the frictional resistance allocated to heat production on the main trace were ≲100 bars. Frictional work allocated to surface energy of new fractures is probably unimportant, and hydrologic convection is not likely to invalidate the conduction assumption, since the heat discharge by thermal springs near the fault is negligible. Explanations for the low dynamic friction fall into two intergradational classes: those in which the fault is weak all of the time and those in which it is weak only during earthquakes (possibly just large ones). The first class includes faults containing anomalously weak gouge materials and faults containing materials with normal frictional properties under near-litho-static steady state fluid pressures. In the second class, weakening is caused by the event (for example, a thermally induced increase in fluid pressure, dehydration of clay minerals, or acoustic fluidzation). In this class, unlike the first, the average strength and ambient tectonic shear stress may be large, ~1 kbar, but the stress allocated to elastic radiation (the apparent stress) must be of similar magnitude, an apparent contradiction with seismic estimates. Unless seismic radiation is underestimated for large earthquakes, it is diffucult to justify average tectonic stresses on the main trace of the San Andreas fault in excess of ~200 bars. The development of the broad Coast Range heat flow anomaly southward from Cape Mendocino suggests that heat flow increases by a factor of 2 with 4 m.y. after the passage of the Mendocinco Triple Junction. This passage leaves the San Andreas transform fault zone in its wake; the depth of the anomalous sources cannot be much greater than the depth of the seismogenic layer. Some of the anomalous heat may be supplied by conduction from the warmer mantle that must occur south of the Mendocino transform (where there is no subducting slab), and some might be supplied by shear heating in the fault zone. With no contribution from shear heating, extreme mantle upwelling would be required, and asthenosphere conditions should exist today at depths of only ~20 km in the northernmost Coast Ranges. If there is an appreciable contribution from shear heating, the heat flow constraint implies that the seismogenic layer is partially decoupled at its base and that the basal traction is in the sense that resists right lateral motion on the fault(s). As a result of these basal tractions, the average shearing stress in the seismogenic layer would increase with distance from the main fault, and the seismogenic layer would offer substantial resistance to plate motion even though resistance on the main fault might be negligible. These speculative models have testable consequences.