Multi-axial stationary non-Gaussian random vibration control tests can simultaneously control the time-frequency charac? teristics of the specified response signals. A fast method for generating stationary non-Gaussian random vibration signals with speci? fied power spectral density， skewness and kurtosis is proposed. The target power spectral density is designed as a filter by frequen? cy sampling method. The non-Gaussian random signal is obtained by nonlinear transformation method and then it passes through the designed filter to obtain the desired non-Gaussian random signals. This method is computationally efficient and overcomes the shortcoming of the traditional nonlinear transformation methods. This proposed method is applied to the three-axis stationary nonGaussian random vibration test， and the closed-loop equalization step of the three-axis non-Gaussian random vibration control is given， which can decouple the power spectrum auto spectrum， coherence coefficient， phase difference， skewness and kurtosis of the signal at the same time. A three-axis stationary non-Gaussian random vibration control test is carried out. The control results of power spectral density， skewness and kurtosis in three directions are satisfactory which meet the requirements of engineering appli? cation.