Embedded zero-tree wavelet encoding (EZW) algorithm is one of the well-known most efficient wavelet image encoding algorithms. In the EZW algorithm, different wavelet decomposition levels will affect the time-frequency resolution of the wavelet as well as the variation of the wavelet coefficients, therefore affecting the final encoding efficiency. Moreover, the times of scan will directly affect the wavelet coefficients to be encoded, therefore affecting the final encoding gain. In this paper, the effects of the wavelet decomposition levels and the times of the scan are investigated and discussed, therefore providing the instruction for the choices of the wavelet decomposition level and the times of scan in the EZW algorithm. From the experiments, we can see that in the same times of scan, with the increase of the wavelet decomposition levels, the image compression ratio (CR) increased accordingly, while the peak signal-to-noise ratio (PSNR) decreases. In the same wavelet decomposition level, with the increase of times of scan, the PSNR increases greatly while the CR decreases.