The 1989 6.9 Loma Prieta earthquake is the first major event to occur along the San Andreas fault (SAF) zone in central California since the 1906 M 7.9 San Francisco earthquake. Given the complexity of this event, uncertainty has persisted as to whether this earthquake ruptured the SAF itself or a secondary fault. Recent work on the SAF in the Coachella Valley in southern California has revealed similar complexity, arising from a nonplanar, nonvertical fault geometry, and has led us to reexamine the Loma Prieta event. We have compiled data sets and data analyses in the vicinity of the Loma Prieta earthquake, including the 3D seismic velocity model and aftershock relocations of Lin and Thurber (2012), potential field data collected by the U.S. Geological Survey following the earthquake, and seismic refraction and reflection data from the 1991 profile of Catchings et al. (2004). The velocity model and aftershock relocations of Lin and Thurber (2012) reveal a geometry for the SAF that appears similar to that in the Coachella Valley (although rotated 180°): at Loma Prieta the fault dips steeply near the surface and curves with depth to join the moderately southwest‐dipping main rupture below 6 km depth, itself also nonplanar. The SAF is a clear velocity boundary, with higher velocities on the northeast, attributable to Mesozoic accretionary and other rocks, and lower velocities on the southwest, attributable to Cenozoic sedimentary and volcanic rocks of the La Honda block. Rocks of the La Honda block have been offset right‐laterally hundreds of kilometers from similar rocks in the southern San Joaquin Valley and vicinity, providing evidence that the curved northeast fault boundary of this block is the plate boundary. Thus, we interpret that the Loma Prieta earthquake occurred on the SAF and not on a secondary fault.