Codes for 3-D electromagnetic (EM) modeling are limited by hardware and time constraints to models that only crudely resemble the geological structures encountered in mineral exploration. They are many orders of magnitude away from execution speeds that would allow routine inversion for models of realistic size and complexity. Nevertheless, important advances in the next few years will come from approximate 3-D methods. Examples are inversion of airborne and reconnaissance ground data in the inductive limit (high-frequency approximation), inversion by an EM analog of seismic migration, and modeling and inversion by fast algorithms allowing multiple plates in a conductive earth in the resistive limit (low-frequency approximation). Full 3-D modeling will be necessary for the development and validation of these approximations.