The combination of the specificity provided by fluorescence microscopy and the ability to quantitatively analyze specimens in three dimensions allows the fundamental organization of cells to be probed as never before. Key features in this emergent technology have been the development of a wide variety of fluorescent dyes or fluorescently labeled probes to provide the requisite specificity. High-quality, cooled charge-coupled devices have recently become available. Functioning as nearly ideal imagers or "electronic film," they are more sensitive than photomultipliers and provide extraordinarily accurate direct digital readout from the microscope. Not only is this precision crucial for accurate quantitative imaging such as that required for the ratioing necessary to determine intracellular ion concentrations, but it also opens the way for sophisticated image processing. It is important to realize that image processing isn't simply a means to improve image aesthetics, but can directly provide new, biologically important information. The impact of modern video microscopy techniques (Allen, 1985; Inoué, 1986) attests to the fact that many biologically relevant phenomena take place at the limits of conventional microscopy. Image processing can be used to substantially enhance the resolution and contrast obtainable in two dimensions, enabling the invisible to be seen and quantitated. Cells are intrinsically three-dimensional. This can simply be a nuisance because of limited depth of focus of the microscope or it could be a fundamental aspect of the problem being studied. In either case, image processing techniques can be used to rapidly provide the desired representation of the data. In this chapter we have discussed the nature of image formation in three dimensions and dealt with several means to remove contaminating out-of-focus information. The most straightforward of these methods uses only information from adjacent focal planes to correct the central one. This approach can be readily applied to virtually any problem and with most commonly available image processing hardware to provide a substantially deblurred image in almost real time. In addition to covering more sophisticated algorithms where the utmost in three-dimensional imaging is required, we have developed a method for extremely rapidly and accurately producing an in-focus, high-resolution "synthetic projection" image from a thick specimen. This is equivalent to that produced by a microscope having the impossible combination of a high-NA objective lens and an infinite depth of focus. A variation on this method allows efficient calculation of stereo pairs.(ABSTRACT TRUNCATED AT 400 WORDS)