FEA(finite element analysis) makes it possible to predict the performance of a structure in the design step. But an accurate FE model is essential to have a reliable prediction. An accurate FE model means that the result from FE model is very close to that from experiment. Therefore, there have been many works to raise the reliability of FE model by modifying it with experimentally measured data. There are two kinds of methods to modify FE model depending on the used experimental data. One is using modal model and the other is using FRF directly. Also the selection of FRF measuring DOFs considerably affects on the FE model tuning. In this work, the modifications of FE model using modal model and FRFs are studied and compared. After choosing measurement DOFs with the degree selection algorithm, the FRFs are measured at these points and used for FE model tuning. The result from using these FRFs are compared with that using modal model, and it is shown that using FRFs can lead to a better result than using modal model, because of the inherent truncation errors in modal model. The effectiveness of the selection algorithm is also shown by comparing the FE model tuning results from the FRF measurements at the best selection DOFs with those from the arbitrarily chosen DOFs. And it is recommended that in order to let the modified FE model remain physically meaningful, the number of design variables shouldn't be larger than that of the frequencies at which the FRFs are compared with each other.