A Raman spectroscopy denoising method based on complete ensemble empirical mode decomposition with adaptive noise is proposed. Permutation entropy is employed to judge intrinsic mode functions (IMFs), if they are on behalf of noise. The IMFs should be filtered based on threshold value. Then the reconstructed signal is without noise. This method is used to denoise the Raman spectra of ethanol solution. The result shows that the method can effectively remove noise from Raman spectroscopy. Compared with empirical mode decomposition and ensemble empirical mode decomposition, the method does not only achieve better performance at high signal to noise ratio (SNR), but also has obvious superiority at low SNR.