Concrete structures show time-dependent behavior due to creep and shrinkage of concrete and the uncertainties of creep and shrinkage are very huge. To reduce uncertainties of creep and shrinkage, it is substantially necessary to perform the long-term creep and shrinkage tests, but actual construction process doesn't allow it due to the limited time. Even though the tests are performed in laboratory, the values obtained from the tests could be different from the actual values in construction site because of the different environment between the laboratory and construction site and the model uncertainty itself. It is difficult to predict the long-term behaviors of concrete structures properly if the assumed creep coefficient obtained from Codes or the results of experiments is different from the real characteristics of concrete creep. In this study, for predicting the long-term behavior, the creep coefficients in reinforced concrete beams are estimated using creep sensitivity analysis from the measured deflections with time. And estimated creep coefficients using creep models of ACI Committee 209 and CEB-FIP MC90 are compared.