This article proposes a lossy compression algorithm and scalable multispectral image coding—including blue, green, red, and near-infrared wavelengths—aimed at increasing image quality based on the amount of data received. The algorithm is based on wavelet subband coding and quantization, predictive multispectral image coding at different wavelengths, and the Huffman coding. The methodology was selected due to small satellites’ low data rate and a brief line of sight to earth stations. The test image database was made from the PeruSat-1 and LANDSAT 8 satellites in order to have different spatial resolutions. The proposed method was compared with the SPIHT, EZW, and STW techniques and subsequently submitted to a peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) evaluation; it showed better efficiency and reached compression ratios of 20, with a PSNR of 30 and an SSIM of approximately 0.8, depending on the multispectral image wavelength.