The widely held perception that the San Andreas fault (SAF) is vertical or steeply dipping in most places in southern California may not be correct. From studies of potential‐field data, active‐source imaging, and seismicity, the dip of the SAF is significantly nonvertical in many locations. The direction of dip appears to change in a systematic way through the Transverse Ranges: moderately southwest (55°–75°) in the western bend of the SAF in the Transverse Ranges (Big Bend); vertical to steep in the Mojave Desert; and moderately northeast (37°–65°) in a region extending from San Bernardino to the Salton Sea, spanning the eastern bend of the SAF in the Transverse Ranges. The shape of the modeled SAF is crudely that of a propeller. If confirmed by further studies, the geometry of the modeled SAF would have important implications for tectonics and strong ground motions from SAF earthquakes. The SAF can be traced or projected through the crust to the north side of a well documented high‐velocity body (HVB) in the upper mantle beneath the Transverse Ranges. The north side of this HVB may be an extension of the plate boundary into the mantle, and the HVB would appear to be part of the Pacific plate.