Three-dimensional images can be assembled by piling up consecutive confocal fluorescent images obtained by confocal scanning laser microscopy. The present work was based on three-dimensional (50-microns-deep) images at high (x, y) resolution obtained with an MRC-500 after en bloc staining of thick slices of rat liver by chromomycin A3 for nuclear DNA. The results of studies on bleaching, fluorescence excitation and emission intensities at various depths of histologic preparations are described. These effects could be evaluated separately by acquiring piled-up ("brick-stepping") and non-piled-up ("side-stepping") (x, y) images at consecutive depths and also (x, z) images. Empirical equations allowed the fitting of experimental plots of bleaching versus time, at different laser intensities and at different depths, and of fluorescence emission intensity versus depth. The main conclusions were that under our experimental conditions: (1) there was no attenuation by depth of the fluorochrome penetration, (2) there was no attenuation of the exciting beam intensity up to at least 50 microns deep, (3) there was an attenuation of the fluorescence emission intensity by depth, (4) bleaching happened equally on all planes above and below any confocal plane being studied, and (5) the fluorescence bleaching half-life was independent of depth. A mathematical correction scheme designed to compensate for bleaching and for attenuation of fluorescence emission in depth is presented. This correction is required for obtaining three-dimensional images of better quality, for optimal three-dimensional image segmentation and for any quantitative analysis based upon voxel-discretized emission intensities (gray levels)--e.g., estimating, by confocal image cytometry, textural chromatin parameters and nuclear DNA amounts.