Various models have been developed to address the problem of "Fermi level pinning," i.e., why the barrier height varies much less than the Schottky metal work function limit. The most widely accepted mechanism is some variant of the metal-induced gap state model. However, recent experimental data on (100)In xGa1-x As surfaces and interfaces (0≤x≤1) suggest that the surface or interface Fermi level can assume values which lie well outside the variance associated with Fermi level pinning; in fact, recent data suggest the achievement of the Schottky limit. Furthermore, studies of epitaxially grown layers where dopant incorporation is dependent on an interface Fermi level suggest that such Fermi level positions are not always pinned. In order to account for these recent results, along with the pinned values, we introduce the concept of an insulating layer like surface reconstruction. Recent calculations suggest that reconstructions of the GaAs(100) surface are insulating. Thus, we suggest that the GaAs(100) metal interface is often a metal (M)/surface reconstruction layer (I)/GaAs bulk (S), or MIS-like. This approach attempts to reconcile disparate models of interface behavior by showing the limits of validity of these models with respect to the actual physical structure of the interface in question.