An extensive review is presented of the various factors involved in the standardization of dyes and stains for cytologic material, which is a crucial point in achieving objective and reproducible measurements of cells for recognition by automated microscopy (high-resolution cell image analysis; automated cell pattern recognition [ACPR]). Two principal methods of standardization are considered at length: (1) standardization based on the physical and chemical characterization of the dyes (physicochemical standardization), as adopted by the German Bureau of Standards, and (2) standardization based on direct visual assessment of stain performance (performance standardization), as adopted by the United States Biological Stain Commission. The procedures, achievements and problems of both methodologies are discussed, as are such related matters as staining time, solution pH, etc., especially with regard to the two most important stains in diagnostic cytology: the Papanicolaou stain and the Romanowsky-Giemsa stain. The standardization of these two stains is considered in detail, with an eye towards their suitability for ACPR. The mechanisms of the staining results produced by these stains are examined, as are the extant problems with each. The Papanicolaou stain, while neither standardized nor stoichiometric, has proven to be of use in cytophotometry. Similarly, the Romanowsky-Giemsa stain, while standardized, is also not a stoichiometric stain. Yet both have been successfully used in some aspects of ACPR; no "ideal" stain has yet been found that would make cell samples ideally suited for machine evaluation. It is concluded that the standardization of biologic dyes and stains, which can contribute to the success of ACPR, should be undertaken by multidisciplinary expert panels, in which the current concerned organizations could play a role. It is also concluded that ACPR may, in fact, contribute to the standardization of dyes and stains: that computer-directed morphometry and automated image analysis with sophisticated statistical analysis may become major tools in the validation of data on staining. For example, computer analysis could be used to pinpoint the particular variant of the Papanicolaou stain that leads to the best overall reproducibility of cell descriptors, in effect achieving "preliminary standardization by computer judgment of stain performance."