In order to get more accurate modeling or forecasting results, a Bayesian multi-model comprehensive framework-IBUNE (Integrated Bayesian Uncertainty Estimator) was used in this paper to analyze the uncertainties from observation data, model parameter and structure based on probability and statistics methods. Research based on Bayesian theory, SCEM-UA algorithm and EM algorithm was embedded in XAJ model and TOPMODEL to optimize parameters and average models. Example shows that IBUNE method can efficiently estimate the uncertainty of hydrologic model, and give a reasonable interval of probability forecast.