Although the importance of “knock” in limiting the power output and efficiency of the spark-ignition engine was realized long ago, and much empirical information has been accumulated which has enabled engine design and fuel quality to be improved—so minimizing the restrictions due to knock—progress in understanding the fundamental nature of the knocking process has been comparatively slow. Recent work, both in Great Britain and America, has improved our knowledge of the chemical and physical factors involved, resulting in a better understanding of the phenomenon. The use of an electromagnetically operated gas sampling valve has enabled the sequence of chemical reactions, occurring in the engine cylinder, prior to knock, to be followed. Organic peroxides have been shown to be some of the most important intermediate products of combustion, and the effect of changing engine conditions—such as compression ratio, speed, and mixture strength—on their formation has been explored. These experiments have shown that there are at least two types of precombustion knock mechanism, one a “low”-temperature two-stage process involving the formation of peroxides, and the other, a “high”-temperature process which is non-peroxidic in character. Studies of the combustion reactions in a motored engine have confirmed these conclusions. So far as the knocking process itself is concerned, there have been two main schools of thought, one that it was an autoignition (i.e., spontaneous ignition) pure and simple and the other that it was a detonation wave perhaps preceded by autoignition. In reviewing published work, the authors consider that the bulk of the evidence favours the autoignition theory. An alternative explanation of the N.A.C.A. high-speed photographic work is put forward which would appear to support the conception of a two-stage autoignition.