The loudness of powertrain noise generally increases with increasing rpm. In the case of ‘linear’ powertrain acceleration sound, the loudness versus time relationship is well described by a linear function. Two studies were conducted on powertrain linearity. The first used tests of similarity and preference to determine whether subjects could detect changes in linearity. The second used a subjective test of preference to investigate how subjects' preference varied with differing degrees of linearity. In both studies, stimulus sets were created by artificially introducing a controlled degree of non-linearity into a nominally linear powertrain sound. The results of the first study indicate that linearity is a phenomenon that naive subjects can readily detect, and that it has an effect on overall preference. Furthermore, the second study shows that preference is related to the magnitude and position of nonlinearities in the growth of loudness versus time during an acceleration run. Based on these results, an objective model is developed that predicts the effects of powertrain linearity on subjects' preference.