Abstract: The sensitivity of the simulated MJO to different cumulus parameterization schemes employed is studied by using the atmosphere general circulation model-SAMIL-R42L9 developed by the Laboratory for Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG) in the Institute of Atmospheric Physics (IAP) of Chinese Academy Sciences. Results show clearly that the performance of the model in simulating the MJO changed widely when using two different cumulus parameterization schemes——the moist convective adjustment scheme (MCA) and the Zhang-McFarlane scheme, respectively. The MJO simulated by the MCA scheme was found to be more realistic than that by the Zhang-McFarlane- scheme including the intraseasonal timescales and eastward propagation. The MJO produced by the Zhang-McFarlane scheme was too weak and showed little propagation characteristics. The weak moisture convergence at the low level simulated by the Zhang-McFarlane scheme is not strong enough to maintain the structure and the eastward propagation of the oscillation. This two cumulus schemes produced different vertical structures of the heating profiles. The heating profile in the ZhangMcFarlane scheme is nearly uniform with height and the heating strength is too weak compared to that produced by the MCA scheme, which maybe contributes greatly to the failure of simulating reasonable MJO. Comparing simulated MJOs by the two schemes indicate that the intraseasonal oscillations simulated by the general circulation model are highly sensitive to cumulus parameterization schemes employed. The performance of the general circulation model in simulating MJO can change widely when the cumulus parameterization scheme is changed. For different cumulus parameterization schemes, the diabatic heating profiles produced by them plays important role on the performance of the GCM. In order to argue further the impacts of the diabatic heating profiles on the MJO simulation, three sensitive experiments with different heating profiles were designed in which modified heating profiles respectively peak in the upper (UH), middle (MH), and lower troposphere (LH). Both the LH run and MH run produced eastward propagating signals at the intraseasonal timescales, while it is interesting to note that the intraseasonal timescales signals produced by the UH run propagated westward. This indicates that realistic intraseasonal oscillation is more prone to be excited when the maximum heating concentrates in the middle-low level, especially in middle level, while, westward propagating disturbances on the intraseasonal time scales are more prone to be produced when the height of the maximum heating is too high.