Spectral unmixing refers to the process by which the spectrum measured over a mixed target is decomposed into a collection of pure component spectra called endmembers and their fractional abundances. Mixed targets are frequently encountered in remote sensing and spectral unmixing techniques have been mostly developed for visible and near-infrared (400'1000nm) and short-wave infrared (1000'2500nm) remote sensing. The first unmixing studies were based on the implicit assumption that the spatial distribution of endmembers is random, and therefore, each mixed pixel is spatially independent. Hence, only the spectral dimension was taken into account in the unmixing process. Nevertheless, a spatial autocorrelation exists between pixels, reflecting the tendency of neighboring pixels to present similar characteristics. Considering these aspects, some developments have been initiated to introduce spatial information in the spectral unmixing process, called the spectral'spatial unmixing technique. This chapter aims to provide a review of these developments to the remote sensing community.