The use of falling weight deflectometer—based backcalculation techniques to determine pavement layer moduli is a cost-effective and widely used method for the structural evaluation of an existing pavement. The nonlinear stress-sensitive response of pavement geomaterials has been well established, and mechanistic-based pavement design can be improved by inclusion of these nonlinear material properties. To further the science of nonlinear backcalculation, the TRB Strength and Deformation Characteristics of Pavement Sections Committee has assembled four data sets that can be used to demonstrate the ability to derive stress-dependent moduli for pavement layers. In this study, validated artificial neural network (ANN)—based backcalculation-type flexible pavement analysis models were used to evaluate the TRB Nonlinear Pavement Analysis Project data sets. The Illi-Pave finite element (FE) model, considering nonlinear stress-dependent geomaterials characterization, was utilized to generate a solution database for developing the ANN-based structural models. Such use of ANN models enables the incorporation of needed sophistication in structural analysis, such as FE modeling with proper materials characterization, into routine practical design. This study illustrated the complexities associated with interpreting the backcalculated modulus values. In general, the predicted strains agreed reasonably well with the measured strain values, whereas the predicted stresses did not.