Multispectral remote sensing images are those obtained from one object by the multispectral sensor in multiple narrow spectral ranges. Large amount of multispectral image data produced by high spatial and spectral resolution poses a great challenge to transmit and storage. As a result, it becomes imperative to develop an efficient compression technique. Since the loss of useful information will affect further processing and applications of the images, a lossless scheme is preferred. In this paper an analysis of the multispectral images space and spectral features is presented, and useful lossless compression techniques such as preprocessing, prediction and transformation are introduced, and also a comparative analysis of these techniques is discussed.