Relocation of the Loma Prieta mainshock-aftershock sequence using the JHD technique places the events an average of 1 km to the southwest of the locations determined by other authors. The new locations show a concentration of seismic activity in the wedge between the San Andreas and Sargent faults. Furthermore, there seems to be little seismic activity to the east of the Sargent fault. These features of the seismicity are not observed in other published locations determined using similar data sets. A depth cross section perpendicular to the Sargent fault near its intersection with the San Andreas fault shows two clusters of events along a line segment between 3 and 10 km that dips to the NE and that intersects the main SW-dipping plane. These clusters may correspond to secondary faults postulated by other authors. The JHD station corrections vary between −0.6 and 0.6 sec, and indicate the presence of large velocity variations across and along the San Andreas fault. These corrections are very close to the weighted average of the station residuals obtained when the events are located singly. The locations obtained with a single-event location program and using the average residuals as “station corrections” give locations almost coincident with the JHD locations. This high similarity of locations is very unusual, and is interpreted as the result of quasi-random lateral velocity variations in the Loma Prieta area. To estimate the effect of these variations on the location of events, synthetic travel times were computed for a realistic, although preliminary, three-dimensional (3D) velocity model. These synthetic data were then treated as the actual data. The corresponding results show that the JHD technique is able to locate events with an average error of 0.5 km in epicenter. Depths, on the other hand, are consistently deeper by about 1 km. In addition, the station corrections are close to the observed corrections. Additional tests with synthetic data also show that the observed station corrections suggest velocity variations of up to ±40% in the top 3 km of the crust. Although these variations may seem extremely large, they are within the range of variation found by other authors and, most importantly, are required by the observations.