As seismic processing technologies have advanced, model definitions have grown correspondingly more complex, progressing from isotropy, to transverse isotropy, to the much more general orthorhombic anisotropy. This growth, while supported partly by continued improvements in acquisition techniques and survey geometries, has progressed at such a rate that current projects require solutions to a much greater relative number of unknowns than those in the near past. This added complexity creates ill-posed problems with huge model spaces, leading to a high degree of uncertainty in the final solutions. This uncertainty can be greatly mitigated through pragmatic use of a priori knowledge and intelligent data leverage and model regularization. Orthorhombic anisotropy has been well characterized both microscopically and macroscopically. This understanding allows an unprecedented level of constraint and confirmation for the model-building process by validating the directionality and strength of azimuthal velocity variations. Information within the data itself can also lead to a much more well-determined model-building result. The observed structure can be used to precondition inversion results to ensure geologic plausibility, constraining similar updates to similar events. Concurrently, nonparameterized residual moveout picking allows for complete freedom in describing gather events, yielding results which best resolve the various anisotropic parameters by accurately fitting the gather data and generating a high-resolution model. By properly combining the various constraints available in a well-designed processing project, the myriad uncertainties that arise when moving to an orthorhombic model space can be simplified and reduced, allowing for geologically reasonable solutions that fit data, yield valuable structural information, and enhance interpretation possibilities.