The Monte-Carlo method is used to investigate the effect of random noise on empirical mode decomposition of the nonlinear signals. The simulation results show that, the influence of noise is obvious for the low-level intrinsic mode function and unabvious for the high-level intrinsic mode function. With the increase of the intensity of white noise, the intrinsic mode function pure noise level will increase. When the intrinsic mode function pure noise levels is subtracted from the noise signal, about 80% of the noise influence will be reduced. The noise signals largest Lyapunov exponent is smaller than that of the noise-free signal.