Convection-Allowing Ensemble Prediction System (CAEPS) has obvious advantages in predicting the convective events due to its fruitful probabilistic forecast information. The CAEPS has become one of the hot focuses in researching and developing the local high-resolution numerical weather prediction (NWP) system. Compared with global ensemble prediction system, representation of model uncertainty in CAEPS is lack of systematic research and theoretical basis, and becomes an important issue worthwhile further research. This paper devotes to reviewing the current state of CAEPS and the studies in representing the model uncertainty over the past 10 years. Up to now, several approaches have been developed in representing model uncertainties, including multi-model, multi-physic, multi-parameter and stochastic physics. These approaches have been widely applied in ensemble forecast of severe convective weather, tropical cyclone intensity and tracks and so on, but with limited effect in improving under-dispersion problem of CAEPS. Such limited effect may come from deficiency of these approaches in formulating the model uncertainties related to small-to-meso-scale system. Except for reviewing the past researches, we propose a way to detect and describe model uncertainties at convective scale.