Previous seismological studies have placed the source of the uplift of the Sierra either within the crust, suggesting a Mesozoic age for the source of the uplift, or in the upper mantle, consistent with late Cenozoic creation of the buoyant material producing the uplift of the range. We deployed 16 temporary seismometers in the high part of the southern Sierra Nevada to augment the permanent Southern California Seismic Network and record arrivals from regional and teleseismic earthquakes. Arrival times of P waves from 54 teleseisms recorded at these stations are advanced by over a second by a high-velocity body in the upper mantle west and northwest of Lake Isabella. Inversion of the arrival times indicates that this ''Isabella anomaly'' is of limited north-south extent (about 40--60 km), has compressional velocities about 4--5% higher than its surroundings, and probably extends from about 100 to 200 km depth. The limited north-south extent of the ''Isabella anomaly'' indicates that it is unrelated to the Sierra; we speculate that it is the downgoing part of a small scale convection system similar to that inferred beneath Southern California. This inversion does not clearly reveal either a large crustal root or a substantial low-velocity body in the upper mantle beneath the Sierra. Although the presence of either degrades the fit to the arrival times and requires high-velocity material beneath the low-velocity material of either root, Bouguer gravity anomalies require low-density material under the Sierra. Assuming that arrival times from earthquakes 150--350 km north and south from the southern Sierra come from a common refractor (the one-layer structure), the upper mantle P wave velocity (Pn) beneath the High Sierra is about 7.6--7.65 km/s; if the arrivals from north and south are from different refractors (the two-layer structure), material with a P wave velocity greater than ~7.2 km/s (the ''7.x'' layer) would lie under a nearly flat interface from more normal crustal velocities and be separated by a north-dipping interface from underlying mantle with velocities about 7.9--8.1 km/s.

The Pn velocity beneath the region immediately to the east is significantly greater (7.9--8.0 km/s) than that of material at equal depths under the Sierra. For the one-layer structure, further assuming that mean crustal velocities are uniform along north-south lines, we find little dip on the Moho in the area; using the arrival times from earthquakes to the south, we infer a depth of 33¿5 km for the Moho beneath the southern High Sierra. This structure of a thin to normal crust over a low-velocity mantle can bereconciled with earlier observations that were used to infer a thick crust under the Sierra. By considering the Bouguer gravity anomaly, the surface geology, refraction profiles in this region, and our own observations, we suggest that 1/3 of 1/2 of the modern elevation in the range is supported by lateral (east-west) density contrasts in the crust; the remainder is supported by density contrasts in the uppermost mantle or lateral variations in the thickness of the ''7.x'' layer. Our interpretation is that the southern Sierra overlies mantle lithosphere that has been thinned and warmed in response to regional lithospheric extension in Neogene time. This part of the upper mantle might have provided the melt that migrated to the east and produced volcanics in the southwestern Great Basin; depletion of the upper mantle might have increased the seismic velocity and decreased the density of material about 60--100 km beneath the southern Sierra.