In 1999, the U.S. Geological Survey and the Southern California Earthquake Center (SCEC) collected refraction and low-fold reflection data along a 150-km-long corridor extending from the Santa Monica Mountains northward to the Sierra Nevada. This profile was part of the second phase of the Los Angeles Region Seismic Experiment (LARSE II). Chief imaging targets included sedimentary basins beneath the San Fernando and Santa Clarita Valleys and the deep structure of major faults along the transect, including causative faults for the 1971 M 6.7 San Fernando and 1994 M 6.7 Northridge earthquakes, the San Gabriel Fault, and the San Andreas Fault. Tomographic modeling of first arrivals using the methods of Hole (1992) and Lutter et al. (1999) produces velocity models that are similar to each other and are well resolved to depths of 5-7.5 km. These models, together with oil-test well data and independent forward modeling of LARSE II refraction data, suggest that regions of relatively low velocity and high velocity gradient in the San Fernando Valley and the northern Santa Clarita Valley (north of the San Gabriel Fault) correspond to Cenozoic sedimentary basin fill and reach maximum depths along the profile of ???4.3 km and >3 km , respectively. The Antelope Valley, within the western Mojave Desert, is also underlain by low-velocity, high-gradient sedimentary fill to an interpreted maximum depth of ???2.4 km. Below depths of ???2 km, velocities of basement rocks in the Santa Monica Mountains and the central Transverse Ranges vary between 5.5 and 6.0 km/sec, but in the Mojave Desert, basement rocks vary in velocity between 5.25 and 6.25 km/sec. The San Andreas Fault separates differing velocity structures of the central Transverse Ranges and Mojave Desert. A weak low-velocity zone is centered approximately on the north-dipping aftershock zone of the 1971 San Fernando earthquake and possibly along the deep projection of the San Gabriel Fault. Modeling of gravity data, using densities inferred from the velocity model, indicates that different velocity-density relationships hold for both sedimentary and basement rocks as one crosses the San Andreas Fault. The LARSE II velocity model can now be used to improve the SCEC Community Velocity Model, which is used to calculate seismic amplitudes for large scenario earthquakes.