Vehicles can experience impacts due to harsh road conditions. Contact with an uneven road surface causes vehicles to vibrate, which generates a loud impact sound. The attenuation of such noise is important because car passengers may complain about the impact noise. However, perfect removal of impact noise is not possible because most of it is caused by external conditions. More research is needed on the objective attributes of impact noise; however, the problem of impact noise is not a simple matter because impact noise is transient in nature and reaches a high level instantaneously. In this paper, a new objective attribute of impact noise is designed using the wavelet transform, which is appropriate for analyzing nonstationary signals, such as an impact signal. The usefulness of the new objective attribute, which is a sound metric, is examined by comparing the mean subjective ratings for real impact noise in passenger cars. The new sound metric has better correlation with the mean subjective rating than currently existing sound metrics.