Conventional friction models predict a substantial thermal anomaly associated with active traces of strike-slip faults, but no such anomaly is observed from over 100 heat-flow determinations along 1,000 km of the San Andreas fault. The Cajon Pass well is being drilled to bring deep heat-flow and stress data to bear on this paradox. Preliminary heat flows to 1.7 km in the neighboring Arkoma well indicated a positive anomaly of about 30% as would be expected if frictional resistance of the fault were 50 MPa or so, but the Pliocene and Pleistocene history of the Transverse Ranges, particularly high rates of uplift and erosion, could also explain the anomaly. Preliminary stress results from Cajon Pass and a new interpretation of regional data by Mark D. Zoback and colleagues suggests that the maximum compressive stress near the fault is almost normal to the trace, and hence the resolved shear stress is low and the fault, weak. Although this is consistent with long-standing evidence from low heat flow, vexing questions remain regarding the physics of strike-slip faulting. The heat-flow data show large variability with depth, probably from three-dimensional structure, and an overall decrease from over 90 mW/m² in the upper kilometer to less than 80 mW/m² in the lower 300 m with no evidence of advective heat transfer. Whether this decreasing trend is significant and whether it continues downward to the background value (∼70 mW/m² less the radiogenic heat from overlying rocks) as expected for a weak fault must be verified with data from greater depth.