The possible development of second-harmonic generation as a tool for studying buried interfaces depends on our ability to understand its dependence on the electronic structure. When the bulk crystals have inversion symmetry, the only dipole-allowed response involves the interface; however, the analysis is much more complicated than with small molecules, and we must develop a qualitative theory of the relevant physics. Here, I propose a closed-form expression for the intensity and apply it to a simple model based essentially on a combination of Fresnel's formulas and a solvable linear chain. The theory allows one to study the second-harmonic spectrum as a function of substrate band structure and plasma frequency, overlayer chemical properties, and experimental geometry. The model is very flexible, and can mimic a variety of electronic structures. Here, for the purpose of illustration, I focus on semiconductors covered by a monolayer of adatoms. The results show that a high degree of surface sensitivity is indeed attained and the (generally very weak) intensity is enhanced when the overlayer is chemically unlike the bulk of the crystal. Particularly strong signals are predicted in a range about the plasma frequency, if the overlayer-induced transitions have high enough energy.