AbstractThis study revisits MJO predictability based on the “perfect model” approach with a contemporary model. Experiments are performed to address the reasons for substantial uncertainties in current estimates of MJO predictability, with a focus on the influence of atmospheric convection parameterization. Specifically, two atmospheric convection schemes are applied for experiments with the NOAA Climate Forecast System, version 2 (CFSv2). MJO potential predictability and prediction skill are assessed, with MJO indices taken as the first two principal components of the combined fields of near-equatorially averaged 200-hPa zonal wind, 850-hPa zonal wind, and outgoing longwave radiation at the top of the atmosphere. Analyses indicate that the convection scheme alone can have substantial influence on the estimate of MJO predictability, with estimates differing by as much as 15 days. Further diagnostics suggest that the shorter predictability with one convection scheme is mainly caused by too weak of an MJO signal. The choice of atmospheric convection scheme also exerts effects on the phase dependency of MJO predictability, and the “Maritime Continent prediction barrier” is identified to be more evident with one convection scheme than with the other.