The bombardment of alloys, oxides, and halides often leads to marked compositional changes at the surface, and these changes have been attributed to an interplay of mass-dependent effects, chemical bonding, electronic processes, and diffusion. We attempt here to answer the limited question of whether, considering only alloys and oxides, mass or bonding is normally more important. The relevant theory is reviewed and extended, with mass effects being shown to be associated most explicitly with recoil sputtering and bonding effects being shown to be associated with all three of cascade sputtering, thermal sputtering, and surface segregation. As far as experimental examples are concerned, mass correlations are found to be quite unsuccessful, whereas most observations can be understood rather well in terms of bonding. Nevertheless, there is a basic problem in that the cascade component of sputtering, normally judged to be predominant, should give significantly less compositional change than is observed. Thermal sputtering would lead to more significant changes, but there is a new problem that, at least with alloys, the absolute yields are probably rather small. A combination of surface segregation with sputtering would also lead to more significant changes, but it is unclear whether segregation is rapid enough to be important in room-temperature bombardments. © 1980, All rights reserved.