Coupled THM simulator on FEBEX case is a highly nonlinear system including several nonlinear sub-models. To better understand the coupled THM processes and their influence on the prediction of disposal system behavior, this paper presents an uncertainty study and model optimization on coupled THM simulation under DECOVALEX-THMC framework. It focus on the effect of complexity on model uncertainty which is based on the hypothesis stating that as a model becomes more complex in terms of increased number of processes involved and parameters, the error between simulations and measurements decreases and the overall model sensitivity increases. An uncertainty function is defined which minimizes both error and sensitivity. The result of comparative study basing on different complexity level models of FEBEX case study verifies this hypothesis and indicates that method presented can address uncertainty of numerical model and give us a criterion to choose most suitable model in a relative manner.