Abstract: To detect knock, a detected signal relating to a combustion property in a cylinder of an engine is generated. The detected signal (I) is filtered to select those portions characteristic of knock. However, the filtered signal (I') is only investigated to determine whether knock has occurred, when the unfiltered detected signal (I) meets at least two, and usually four, preset conditions. One preset condition is that the detected signal (I) must have reached a predetermined threshold value (I.sub.s). Other preset conditions for the detected signal include the following: that the maximum value (I.sub.max) of the detected signal (I) is below a predetermined critical value (I.sub.max crit) and that within a predetermined crank angle difference (.DELTA..alpha..sub.F), after reaching the threshold value (I.sub.s), during a predetermined time interval (.DELTA.t) the detected signal (I) does not decrease by more than a first critical amount (.DELTA.I.sub.

Abstract: A process for determining the combustion pressure of an internal combustion engine is proposed. Conventional processes of this type have the disadvantage that the determination methods are very costly. Furthermore, stringent requirements are made of the sensors to be used. By comparison, the process proposed here is distinguished in that the combustion pressure is calculated from the combustion chamber pressure (p(.alpha.)) determined with simple sensors, by determining the compression pressure (p.sub.k (.alpha.)). These values for the compression pressure are subtracted from the combustion chamber pressure in order to calculate the combustion pressure p*(.alpha.)). Preferably, the course of the compression pressure is obtained by measuring the combustion chamber pressure values during the compression until top dead center in the working phase and storing them in a memory.