Every rotating machine exhibits a unique vibration pattern or (signature) which remains constant unless the system parameters or operating conditions change. This vibration pattern can be used for continuous diagnosis of machine performance, detecting and identifying faults as they occur. Consequently, repairs can be made before serious failure takes place. The concept of mechanical fault diagnosis and engine monitoring has been frequently discussed in the recent literature (see, e.g., references [1-7]). Various detection techniques have been developed but in none of the published work has a learning control system been utilized to correct for the engine malfunction. In this paper a report is presented of an investigation of the feasibility of utilizing vibration analysis information as feedback to improve the performance of engines. With a spark ignition engine as a special case, a micropressor-based learning controller and the necessary logic steps are described for detecting change in the engine performance, identifying its cause and attempting to correct it.