The electronic system packaging community has a great need to reduce the size of its heat transfer simulations so that it can: simulate and analyze more complex systems, include additional physical phenomena, and improve its ability to search the electronic systems packaging design space. Aside from further improvements in machine speed and numerical algorithm efficiency, this is basically a question of model reduction and experimental identification: one would like to know how to dramatically reduce the size of heat transfer simulations when they are available, and one would further like to identify models directly from experiment when accurate, computationally feasible, numerical simulations are not available (as in the case of turbulent flows through complex geometries).Fortunately, the topic of low-order modeling for design has been widely studied and successfully applied in other fields (mostly in control engineering and fluid dynamics, although also in structural mechanics, chemical deposition, heat transfer and combustion). Specifically, there are thousands of papers on the mathematical techniques of model reduction and experimental system identification. This paper gives a brief overview of these techniques, it suggests how these tools might be effectively used for electronic systems including cases that involve unsteady fluid dynamics, and it summarizes some of the reduced-order modeling lessons learned in other fields. The paper includes some of our initial work in model reducing the unsteady heat conduction equation, a result on component model inter-connections, and an outline of a systems level model for an air cooled personal computer.