Theoretical predictions of electronic energy levels associated with s- and p-bonded substitutional point defects at (110) surfaces of InAs and other III–V semiconductors are presented and discussed. The specific defects considered for InAs are: anion and cation vacancies, the (native) antisite defects InAs and AsIn, and 26 impurities. The predicted surface-defect deep levels are used to interpret Schottky barrier height data for (a) n- and p-(InAs) and (b) the alloys AlxGa1−xAs, GaAs1−xPx, In1−xGaxP, and In1−xGaxAs. The rather complicated dependence of the Schottky barrier height φB on alloy composition x provides a nontrivial test of the theory (and competing theories). The following unified microscopic picture emerges from these and previous calculations: (1) For most III–V and group IV semiconductors, Fermi-level pinning by native defects can explain the observed Schottky barrier heights. (2) For GaAs, InP, and other III–V semiconductors interfaced with nonreactive metals, the Fermi-level pinning is normally due to antisite defects. (3) When InP is interfaced with a reactive metal, surface P vacancies are created which pin the Fermi level. (4) Impurities and defect complexes are sometimes implicated. (5) At Si/transition-metal-silicide interfaces, Si dangling bonds pin the Fermi level. (6) These defects at the semiconductor/metal interfaces are often ‘‘sheltered’’ or ‘‘encapsulated.’’ That is, the states responsible for Fermi-level pinning are frequently ‘‘dangling-bond’’ states that dangle into a neighboring vacancy, void, or disordered region. The defects are partially surrounded by atoms that are out of resonance with the semiconductor host, causing the defect levels to be deep-level pinned and to have energies that are almost independent of the metal.