Abstract: The GLUE method is an important means of parameter identification under uncertain information. The likelihood function plays an important role in GLUE. The effects of likelihood functions on the parameter identification and model forecast results were discussed with a substance decay model by artificial composite data to avoid the effects of observation errors and model structure errors. From the identification results, the variation of likelihood functions changes the comparison of the likelihood values of the parameters, so the results of parameter identification and sensitivity results are both influenced, even the order of the global sensitivity of the parameters may be different. From the forecast results, the variation of the likelihood functions influences the distribution of themodeling results. On the other hand, although the illegibility of forecast results is decreasing, theaccuracy may not increase at the same time. Because of the important effects oflikelihood functions on the analysis results, this functions should be determinedon the basis of the detailed characteristics of studied results and the requests of the modeling by the probability theory and other methods to reflect the real importance of the parameters.