The homogeneous charge compression ignition (HCCI) and reactivity controlled compression ignition (RCCI) engines are very sensitive to the incylinder compression temperature, the spatial distribution of fuel and air, and the component of fuels. Variation of these facts will result in the cyclic variation in IC engine. Based on the needs of cyclic variation research in HCCI engine and RCCI engine, an online cylinder pressure acquisition and real-time combustion analysis system adopting in-cylinder pressure sensor, data acquisition card, photoelectric encoder and LabVIEW software was built. In order to improve the real-time performance of combustion analysis system, only single cycle cylinder pressure was collected and calculated. There were much pressure fluctuation in single cycle cylinder pressure resulting from channel effect and other effect, and a real-time filtering of in-cylinder pressure was adopted to reduce the interference error of single cycle in-cylinder pressure data. The filtering methods based on spectrum analysis, like fast Fourier transformation (FFT), linear interpolation and inverse fast Fourier transform (IFFT), can realize the online adaptive identification and auto pressure smooth for the single cycle in-cylinder pressure, which made the real-time combustion analysis system for HCCI and RCCI mode possible. Then, based on the smoothed cylinder pressure, some important combustion parameters such as the maximum pressure increase rate, combustion heat release rate and in-cylinder peak pressure were calculated. In order to further improve the real-time performance of the combustion analysis system, the producer/consumer operation mode which can achieve the synchronized process of real-time data calculation and fast data storage was used. The improved data sharing ability can improve the real-time ability of the combustion calculation. Different accuracy levels of calculation methods were adopted for different working stages in engine, which can reduce the calculation time for intake, exhaust, compression and expansion stages. In combustion stage, which had the largest calculating workload, the appropriate simplified calculation formulas and formula node operation modules were adopted to improve the real-time performance. Finally, the real-time performance of the combustion analysis system was analyzed and an experimental verification of the combustion analysis system was carried out. 