Fluorescence microscopy is an essential tool for modern biological research. The wide range of available fluorophores and labeling techniques allows the creation of increasingly complex multicolored samples. A reliable separation of the different fluorescence labels is required for analysis and quantitation, but it is complicated by the significant overlap of the emission spectra. This problem can be addressed on the acquisition and the processing side by the use of spectral imaging in conjunction with linear unmixing of the image data. This method allows the reliable separation of even strongly overlapping fluorescence signals and has become an important tool in colocalization and in FRET studies. In this chapter, the microscope techniques available for spectral imaging are presented and the theory of linear unmixing is explained. Possible limitations as well as approaches for image optimization are discussed to help to realize the full potential of this novel method. Biological applications that can be improved by spectral imaging and linear unmixing are presented.