In this work, the possibility to diagnose the valve lashes of an automotive diesel engine using the engine block vibration signal analysis is studied. For this purpose, several signal processing techniques are adopted to extract the valve lash related information from heavily conteminated vibration signal. Both the time and the frequency domain signal are tested to diagnose the valve lash condition. From the many suggested methods to identify the operating valve lash states, it was found that time domain engine block acceleration signal analysis method is the most reliable to diagnose the valve running condition. The valve closing angle and the peak acceleration due to valve closing are chosen to tell the valve lash condition. All the measured engine block acceleration signal are statistically tested to derive indexes which are useful to judge the valve lash of each valve. In conclusion, it was found that the developed technique can be a tool to diagnose the valve condition while in operation.