In order to prevent harmful vibration in ships at the early stage of design, anti-vibration design is required on the basis of accurate estimation methods. To investigate vibration response of superstructure, very large scale of 3-dimensional finite element analysis including hull structure is sometimes conducted, however, significant improvement of accuracy is not yet achieved due to the difficulty in estimating damping. More advanced treatment of damping is necessary to obtain more accurate vibration responses. In the 1st report, a new identification method of damping factor is proposed by calculating dynamic response directly on the basis of Rayleigh's damping [ C ]=α[ M ]+β[ K ]. Then the capability of this method is verified by the measured data of frame structure model. The 2nd report relates that this method is applied to the measured data of a 280, 000DWT VLCC's superstructure and that its ability is confirmed. In this 3rd report, Rayleigh's damping is enhanced to the form of [ C ]=α_s[ M_S ]+α_w[ M_w ]+β[ K ] where α_s and α_w are the identified coefficients corresponding to the mass matrix of structure and the virtual mass matrix of fluid, respectively. To verify the above assumption, following tests and identifications are carried out: 1) Exciter test of model ship is conducted in air condition and in water condition to measure transfer function. Firstly, α_s and β are identified for model ship in air. Using the same α_s and β, transfer function in water is identified by α_w considering that the difference of transfer functions between in air and in water caused by the increase of fluid damping. Simulated transfer functions in water by these three identification coefficients show good agreement with the measured ones. 2) Exciter test of actual VLCC³⁾ is utilized to identify three coefficients. Same α_w as of model ship is taken in the identification on the assumption that the coefficient of fluid damping of VLCC is approximately similar to that of model ship, α_s and β are identified for VLCC in ballast condition. Then transfer functions in ull load condition is estimated by these three parameter obtained in ballast condition, which coincide with the measured data.