The long-standing ''stress/heat flow paradox'' was the primary scientific motivation for the Cajon Pass borehole. For nearly two decades, the absence of a fault-centered heat flow anomaly from measurements to relatively shallow (~200 m) depths had indicated low average shear stresses (≤20 MPa) on the San Andreas fault, while laboratory data on rock strength and in situ stress determinations to about a kilometer had indicated high stress (~100 MPa). Initial results from an unsuccessful 1.7-km-deep oil well at the site gave a high heat flow (~90 mW m-2) consistent with a strong San Andreas fault; however the late Cenozoic geologic history of the Cajon Pass area suggested that the anomalous heat flow was the transient effect of rapid erosion. Theoretical studies predicted that the ~30%surface anomaly would be substantially reduced at depths of 3--5 km. The research borehole reached a total depth of 3.5 km. Below a superficial covering of Tertiary sedimentary rocks, it penetrated gneissic rocks with composition ranging from gabbroic to granodioritic. Core recovery amounted to only about 3% of the total depth, necessitating the use of drill cuttings to characterize thermal conductivity. This, in turn, resulted in much higher uncertainties in average conductivity (¿10--15%) than would have occurred with a continuously cored hole (¿3--5%). From a time series of temperature logs, equilibrium temperature gradients were established over selected intervals of 250--500 m to within 95% confidence limits of 2%. These gradients were combined with harmonic mean thermal conductivities having larger uncertainties to give interval heat flows, which vary systematically from 100¿5 mW m-2 in the uppermost 400 m to 75¿3 mW m-2 in the lowermost 300 m. Thus, at the Cajon Pass site, heat flow is decreasing with depth at a mean rate of more than 7 mW m-2 per kilometer, consistent with a frictionless fault and with theoretical predictions based on local erosional history. ¿ American Geophysical Union 1992