Measuring and quantifying automobile engine impact noises such as piston slap, valve noise, main bearing noise, involves the analysis of non-periodic signals. Signals of this type are usually easy to detect but difficult to quantify. Most of the common methods now used for identifying non-periodic signals involve FFT based frequency domain and time domain averages, but by the nature of engine impact noises they are most likely position-locked signals. In other words, impact noises from an engine are not related to speed but to crank angle. Therefore, it is preferable that postprocessing mechanical impact noise signals of an engine should be carried out in the crank angle domain.This paper describes a simplified approach for statistically quantifying engine impact noises which has been found to be highly efficient and reliable when applied to SI engine mechanical noises. The methodology involves applying a simple signal processing technique without the aid of an encoder (usually encoders or pulse multipliers are required for extracting crank angles). The paper also describes investigations of the effect of engine oil temperature on cold engine noises using this method.