The Magnitude and Distance Amplitude Correction (MDAC; Taylor and Hartse, 1998; Taylor et al., 2002) procedure for correcting regional seismic amplitudes for seismic event identification has been modified to include more realistic earthquake source models and source scaling. In the MDAC2 formulation we generalize the Brune (1970) earthquake source spectrum to use a more physical apparent stress model that can represent non-constant stress-drop scaling. We also event include a parameter that allows for variable P-wave and S-wave comer frequency scaling, imposing some of the constraints of ratio correction techniques (Rodger and Walter, 2002). VeryStable moment magnitude measures (Mayeda et al., 2002) from regional coda wave envelopes that have been tied to independently derived regional seismic moments are incorporated. This eliminates two fitting parameters that were necessary in relating seismic moment to magnitude. The incorporation of Bayesian tomography to replace the assumption of a constant Q0 model is also described. These modifications allow for more flexibility in the MDAC grid-search procedure. The direct tie to regional seismic moment rather than body wave magnitude reduces effects of upper mantle bias on the corrected amplitudes. In this paper, we develop the theory and test the formulation on Nevada Test Site (NTS) data.