Oversimplifying structural modeling in the analysis of the seismic performance of tall buildings introduces variability and biases in the computed seismic response. Engineering demand parameters (EDPs) from coupled soil–structure interaction (SSI) systems can only be realistically determined when nonlinear constitutive soil and structural behaviors are adopted in the numerical formulation. Large-magnitude earthquakes mobilize soil and structural demands that are beyond the elastic response of both materials. This paper aims at evaluating the influence of the structural modeling assumptions in the computed soil response due to seismic excitations using a direct fully coupled nonlinear SSI approach. Numerical analyses are conducted on elastic and nonlinear inelastic-degrading tall building models supported on a mat foundation-to-soil continuum domain using a multiple-yield surface plane-strain constitutive model. The models were subjected to multiple ground motion scenarios representative of broadband frequency content. Large discrepancies occur between linear and nonlinear inelastic tall buildings in terms of interstory drifts, peak horizontal accelerations, structural displacements, hysteretic energy, foundation rotation, and soil settlements. The results show how using nonlinear inelastic-degrading structural models of tall buildings largely affect the computed seismic response of the entire SSI system relative to linear elastic structural modeling approaches. More realistic responses are obtained using nonlinear degrading building models including SSI effects, because energy distribution and tradeoff among both the supporting soils and structure vary significantly as the seismic demands induce stresses and strains in the building beyond the onset of structural yielding.