In the design process of traditional ship shafting, the design quality is generally hard to get guaranteed for the lack of discipline coupling. In this paper, Multidisciplinary Design Optimization (MDO) is innovatively introduced to ensure design quality. Multidisciplinary decomposition of shafting can help to construct the MDO model of ship shafting based on multidisciplinary feasibility method. Then the sub-discipline model of shifting design can be further established, including calibration neutron discipline model, whirling vibration model, and dynamic stiffness of radial oil film bearing model. Collaborative operation is implemented by the multidisciplinary model of shifting to obtain the experimental results. Based on Radial Basis Function (RBF) neural network, the responsive surfaces of variable, bearing load, and support stiffness can be constructed, in the meanwhile the dynamic stiffness decoupling of vibration model can be obtained. Fireworks algorithm is used to establish multidisciplinary optimization of seven-dimensional design variable. The results show that MDO helps improve the quality of shafting alignment and whirling vibration. The work in present paper also provides insight for the future optimization of research methods, design quality, and engineering experiments.