.[1] Seismic tomography combined with waveform modeling constrains the dimensions and melt content of a magma body in the upper crust at Newberry Volcano. We obtain a P-wave tomographic image by combining travel-time data collected in 2008 on a line of densely spaced seismometers with active-source data collected in the 1980s. The tomographic analysis resolves a high-velocity intrusive ring complex surrounding a low-velocity caldera-fill zone at depths above 3 km and a broader high-velocity intrusive complex surrounding a central low-velocity anomaly at greater depths (3-6 km). This second, upper-crustal low-velocity anomaly is poorly resolved and resolution tests indicate that an unrealistically large, low-velocity body representing $60 km 3 of melt could be consistent with the available travel times. The 2008 data exhibit low amplitude first arrivals and an anomalous secondary P wave phase originating beneath the caldera. Two-dimensional finite difference waveform modeling through the tomographic velocity model does not reproduce these observations. To reproduce these phases, we predict waveforms for models that include synthetic low-velocity bodies and test possible magma chamber geometries and properties. Three classes of models produce a transmitted P-phase consistent with the travel time and amplitude of the observed secondary phase and also match the observed lower amplitude first arrivals. These models represent a graded mush region, a crystal-suspension region, and a melt sill above a thin mush region. The three possible magma chamber models comprise a much narrower range of melt volumes (1.6-8.0 km 3 ) than could be constrained by travel-time tomography alone.Citation: Beachly, M. W., E. E. E. Hooft, D. R. Toomey, and G. P. Waite (2012), Upper crustal structure of Newberry Volcano from P-wave tomography and finite difference waveform modeling,