Imaging of the Garlock fault for seismic reflection data tests tectonic models for the Mojave Desert region of southern California, Models developed from geologic and geodetic evidence of fault movement rates disagree on whether the Garlock fault should dip north or south. Such models are further at odds with focal mechanism and dislocation analyses consistent with a vertical strike-slip fault. Our analysis demonstrates a nonvertical, southward dip for the Garlock fault. Consortium for Continental Reflection Profiling Mojave line 5 collected seismic reflection data across the Garlock in Cantil Valley, where the 3-km-deep Cantil pull-apart basin has developed between the southwest and east branches of the fault. We analyze the reflection data for evidence of the fault's structure to 5 km depths. Garlock fault plane reflections on unprocessed shot gathers merge with the first arrival as the source progresses to the surface trace of the east branch, explicitly tracing the Garlock fault as a seismic reflector from depth to the mapped contemporary fault trace. The apparent velocity of these reflections gives a 37¿¿10¿ south dip for the east branch of the fault. A prestack Kirchhoff sum migration images the east branch reflector with a 45¿ south dip. Our migration process fully accounts for the bending of seismic rays through the strong lateral velocity variations in Cantil basin. The south dip of the Garlock's east branch together with basement steps that decrease to the south suggest that Cantil basin developed from a 0.4--1 mm/yr component of extension normal to the fault. Development of the basin by detachment provides a mechanism to widen it as left-lateral motion lengths it, allowing the pull-apart to maintain the globally constant length-to-width ratio of 3. Detachment by the Garlock at Cantil Valley similary requires 0.4--1 mm/yr of dextral strike slip of northwest striking fault such as the Helendal that cuts the Mojave between the Garlock and Pinto Mountain faults. Such motion is consistent with regional tectonic models. ¿American Geophysical Union 1991