We have developed seismic velocity models for the heated rock surrounding a tunnel in Yucca Mountain tuff and compared the results with field data obtained at the Yucca Mountain drift scale test (DST) facility from 1998 to 2002. During that time, the tunnel was heated to replicate the effects of long-term storage of decaying nuclear waste and to study the effects of extreme temperatures on the surrounding rock and groundwater flow. Our velocity models are based on borehole temperature data, thermal models, and laboratory measurements on granite. Comparisons between field and synthetic seismograms show that superheating the rock around the tunnel causes thermally induced variations in P- and S-wave arrival-time separation. Barring out-of-plane reflections, 2D spectral element waveform modeling in the source plane consistently replicates seismic receiver waveforms and classic behavior of pulses reflected from cylinders. Our models constrain the in situ velocity/temperature derivative of the tuff to be approximately per . This velocity change is consistent with thermally induced wavespeed changes in dry rock samples and is lower than expected for water-to-steam conversion in saturated rock. We infer that velocity changes are controlled by thermal expansion and fracturing. Additionally, we have developed an improved method for monitoring tunnel conditions that uses waves diffracted around the tunnel in the region of changing velocity.