A steady-state model has been proposed to predict the oxygen reactive-ion-etching resistance of organosilicon polymers. This model is based on a Silicon material balance and the assumption that a protective Si02 film forms and reaches a steady-state thickness on the surface of the polymer. At steady-state, the rate determining step is sputtering of the SiO2 film. This model predicts that the steady-state etching rate is proportional to the sputtering rate of Si02 and inversely proportional to the mass density of silicon in the polymer. It also predicts that the etching rate is independent of other chemical or physical properties of the material. This model accurately predicts the etching behavior of a silyl novolac polymer over a wide range of etching conditions. Two silyl methacrylates etch at the predicted rate under conditions typical of trilevel processing, but exceed the predicted rate under conditions where the average bombardment energy is lower. Surface analysis shows that the steady-state approximation is not valid for the methacrylates under these etching conditions; the oxide thickness continues to increase with time, even though a constant etching rate is achieved. Polymers with very low silicon content do not etch according to the model but form highly porous oxides that continuously accumulate on the surface of the polymer.