This study presents the development of sensor less rough road detection by using frequency analysis of instantaneous crankshaft rotation speed signal. Combustion diagnosis (misfire detection) is enforced by OBD requirement. Eliminating rough road’s effect on misfire detection is critical for getting robust diagnosis result, especially for crankshaft rotation speed based misfire detection. Although some kind of sensor is often used to identify vehicle driving on rough road, sensor less rough road detection method is still attractive for OEM to reduce cost and some approaches have been developed in past years for the aim   .This paper presents a new rough road detection approach based on crankshaft rotation speed frequency domain analysis. It firstly collects the signal from vehicle driving with misfire on public road and without misfire on rough road. Fluctuation exists in crankshaft rotation speed under such condition. It is processed as vibration signal in this paper. The different vibration energy distribution pattern in frequency domain is compared. A high pass filter is designed to use the difference for rough road detection. Finally an index is calculated from the signal and its filtered version. Once the index exceeds calibrated threshold, vehicle driving on rough road is assumed present. The method had been validated on a MPV vehicle in field test.