Abstract: Simulation of nonstationary processes has become an indispensable tool in engineering. It is not easy for samples of nonstationary stochastic having same statistic characteristics. A new method is proposed for the simulation of nonstationary random processes based on orthogonal Hilbert-Huang transform (OHHT) spectra of their sample observations. Firstly, orthogonal method is introduced to treat with IMF, and the result shows that orthogonal HHT is a universal method avoiding leakage of energy compared with conventional HHT. Secondly, based on OHHT, the average of the Hilbert spectra over the samples is then defined as the Hilbert spectrum of the peocess and used as the target in the simulation of the process. The statistic characteristics fuction of stochastic process is given in the paper. Simulation of earthquake ground motion characteristic of low frequency and simulation of subwary vibration characteristic of high frequency are carried out in the paper. The time history of samples, Fourier amplitude spectrum of samples and response spectra of samples compare well with those of the records. Samples and record of same process have same statistic characteristics.