Powertrain diagnosis has been demanded with growth & complexity of powertrain electronic control system and enforcement of law & regulation in the last decades. In regulation OBD II, requirement of misfire monitoring has been demanded much more strictly. A variety of diagnosis methods for misfire have been proposed and developed, however most of them either depend greatly on special or expensive sensors or suffer from the disturbance of vibration due to non-misfire reasons. One combination of Frequency Domain Analysis and Fuzzy Logic to perform the misfire diagnosis is proposed. It takes full advantage of property of frequency domain analysis and fuzzy logic, providing accurate and robust detection results, without adding additional hardware diagnosis instruments. Fast Fourier Transform is used to perform frequency domain analysis, which can abstract the specific frequency characteristics of misfire signals during overall engine crankshaft speed fluctuations directly and quickly, and fuzzy logic provides good trading off between significance and precision, which can reduce possible false misfire alarm caused by interruption of noise and disturbance of uneven road. The experiment on one four-cylinder engine proves that the method is effective.