AbstractThis article addresses the problem of internal combustion engine misfire detection. Specific attention is paid to misfire detection using crankshaft speed measurements. After an initial analysis of the problem, the applicability of advanced signal processing methods to its solution is discussed. We demonstrate how adaptive filtering techniques can lead to improved misfire detection accuracy. This can lead to reduced mis‐detection as well as false alarm rates. The general concept of estimating and removing periodic, time‐varying components may also be applied to other automotive problems. Copyright © 2004 John Wiley & Sons, Ltd.