Autoregressive-moving average models are developed to represent grinding wheel profiles for different combinations of sampling parameters including the sample interval, the number of observations, and the length of record. Using 46 and 120 grit grinding wheels, the effects of the choice of sample interval and number of observations on the appropriate model form are investigated. Discrimination between models for different grit size grinding wheels is discussed. A new criterion is proposed for the selection of the sample interval, based on observations per grit (OPG), to achieve comparable discrete approximations of the wheels and to maximize discrimination between models of different wheels.