Close-loop feedback combustion control is essential for improving the internal combustion engines to meet the rigorous fuel efficiency demands and emission legislations. A vital part is the combustion sensing technology that diagnoses in-cylinder combustion information promptly, such as using cylinder pressure sensor and ion current measurement. The promptness and fidelity of the diagnostic are particularly important to the potential success of using intra-cycle control for abnormal cycles such as super knocking and misfiring. Many research studies have demonstrated the use of ion-current sensing as feedback signal to control the spark ignition gasoline engines, with the spark gap shared for both ignition and ion-current detection. During the spark glow phase, the sparking current may affect the combustion ion current signal. Moreover, the electrode gap size is optimized for sparking rather than measurement of ion current. For improving fuel efficiency using diluted combustion, the ion current measurement could be useful for engine control. One solution is to use dedicated ion current probes which would require additional space in the engine, and may increase design complexity. In this paper, the authors propose the use of a multi-electrode spark plug for sparking and ion current measurement. This plug takes up the same area as a conventional spark plug. It has three central electrodes - one is used for sparking, and the other two are used for simultaneous and independent ion current measurement. Signal validation tests were performed in a constant volume combustion chamber using optical methods. Subsequently, the performance of the multi-electrode plug was tested on an engine at different operating conditions such as air-fuel ratio, exhaust gas recirculation, spark timing, and compression ratio. Comparisons were made with a conventional spark plug. It was found that the multi-electrode plug could detect combustion reliably even at dilute conditions. The ability to measure at two locations simultaneously improved the accuracy of the measurement, and minimized false detection.