A “CARS” System using a modeless dye laser has been extensively calibrated and shown to give average temperatures of acceptably good accuracy. It has been used to measure temperatures in the end-gas of a single-cylinder E6 engine under knocking conditions using propane, commercial butane, iso-octane and a primary reference fuel made up of 90% iso-octane and 10% n-heptane by volume. These measurements show that there is significant heating of the end-gas because of pre-flame chemical reactions for all the fuels except propane. Propane has to be compressed to a much higher pressure compared to the other fuels studied in order to make it knock. At a given engine operating condition, there is significant cycle-by-cycle variation in both combustion and knock. For a given fuel and spark timing, at a fixed crank angle, the difference in temperature and pressure between knocking and non-knocking cycles is small early in the cycle but increases at more advanced crank angles; the knocking cycles generally having higher pressures and temperatures. However, because of the better repeatability of the pressure measurements, this difference can be seen in the mean pressures with statistical significance even early on in the cycle. Higher peak pressures are associated with higher knock intensities as measured by the amplitude of the high-frequency pressure fluctuations. Broadly, the earlier the onset of knock, the higher is the knock intensity. Knock events which give similar intensities and peak pressures can start from different onset pressures, possibly indicating the multi-centred nature of the autoignition events preceding knock. Knock events starting at comparable pressures can give rise to significantly different knock intensities. For a given operating condition, for all the fuels studied, faster combustion as measured by the flame transit time from spark breakdown to flame arrival at an ion gap, was associated with higher knock intensity because of the higher pressures generated.