The imaging of living cells and tissues using laser-scanning microscopy is offering dramatic insights into the spatial and temporal controls of biological processes. The availability of genetically encoded labels such as green fluorescent protein (GFP) offers unique opportunities by which to trace cell movements, cell signaling or gene expression dynamically in developing embryos. Two-photon laser scanning microscopy (TPLSM) is ideally suited to imaging cells in vivo due to its deeper tissue penetration and reduced phototoxicity; however, in TPLSM the excitation and emission spectra of GFP and its color variants [e.g., CyanFP (CFP); yellowFP (YFP)] are insufficiently distinct to be uniquely imaged by conventional means. To surmount such difficulties, we have combined the technologies of TPLSM and imaging spectroscopy to unambiguously identify CFP, GFP, YFP, and redFP (RFP) as well as conventional dyes, and have tested the approach in cell lines. In our approach, a liquid crystal tunable filter was used to collect the emission spectrum of each pixel within the TPLSM image. Based on the fluorescent emission spectra, supervised classification and linear unmixing analysis algorithms were used to identify the nature and relative amounts of the fluorescent proteins expressed in the cells. In a most extreme case, we have used the approach to separate GFP and fluorescein, separated by only 7 nm, and appear somewhat indistinguishable by conventional techniques. This approach offers the needed ability to concurrently image multiple colored, spectrally overlapping marker proteins within living cells.