In 1982 and 1984 the U.S. Geological Survey used several seismic networks, totaling over 90 stations, to record teleseismic P waves and measure travel time residuals in an area centered on the Long Valley caldera. We inverted the travel time residuals to obtain a three-dimensional image of the velocity structure with resolution of 5–6 km to depths of 70 km beneath the array. Direct inversion of these data indicates that the 2- to 4-km-thick low-velocity caldera fill contaminates the signal from any midcrustal velocity anomalies beneath the caldera. Thus two methods were used to strip the effects of the upper crust from the travel time residuals: (1) ray tracing through upper crustal velocity models provided by seismic refraction experiments and gravity surveys, and (2) an iterative stripping scheme using the inversion itself. The methods produce essentially identical results and adequately remove the effects of the shallowest crustal structures, including the caldera fill and hydrothermal alteration effects. The resulting “stripped” models show two well-resolved midcrustal low-velocity bodies in the Long Valley region. The first body is centered between 7 and 20 km depth beneath the resurgent dome of the Long Valley caldera and has a volume of 150–600 km³. The second, with a similar volume, is centered between 10 and 20 km depth beneath the Mono Craters, about 10 km north of Long Valley. Velocity contrasts in both of these bodies are about 6–10%, and the features are interpreted as silicic magma chambers. This experiment does not preclude the presence of additional pockets of magma smaller than 5 km across in the upper crust, particularly beneath the resurgent dome of the caldera (which would be removed with the stripping methods). It is likely that the midcrust anomaly beneath the resurgent dome is a remnant of the caldera forming magma chamber. Upper mantle velocities are lower than average beneath Mono Craters and higher than average beneath Long Valley. We tentatively interpret these anomalies as mantle partial melt beneath the Mono Craters and refractory upper mantle residuum beneath Long Valley. The high eruptive rate of the Mono Craters and these upper mantle structures suggest that the focus of volcanism is shifting north from Long Valley to the Mono Craters.