Based on the analyses of environmental satellite multispectral images, a new compression algorithm is proposed based on the three-dimentional set partitioning in hierarchical trees (3D-SPIHT) and region of interest (ROI) coding. To reduce the redundancies in the spatial and spectral domain and decrease the reconstructed spectrum error, a three-dimentional discrete wavelet transformation (3D-DWT) which combines two wavelet bases in the spectral domain is carried out. Then, a method which unites partial 3D-SPIHT with lifting the wavelet coefficients of ROI coding algorithm, is adopted to allocate the coding rate in reason according to the different significance of the reconstructed spectrum in the spatial orientation tree wavelet coefficient. The new algorithm on the probability of the significant coefficients in each bit plane to select one of three different coding modes, which enhances the coding efficiency. Finally, the experimental results show that the proposed algorithm achieves improved performance over the conventional algorithm. With the 8∶1 compression ratio, the algorithm satisfies the requirement of the satellite multispectral image system.