The experimental observations of metallurgical interactions between compound semiconductor substrates and metallic or oxide overlayers have stimulated a new model of Fermi level ’’pinning’’ at these interfaces. This model assumes the standard Schottky picture of interface band alignment, but that the interface phases involved are not the pure metal or oxide normally assumed by other models. For both III‐V and II‐VI compounds, the barrier height to gold is found to correlate well with the anion work function, suggesting the interface phases are often anion rich. This correlation holds even for cases in which the ’’common anion rule’’ fails, and explains both successes and failures of this earlier model.