Models of dental caries (laboratory, animal, and human in situ models) vary markedly in their microbiological complexity. Laboratory models range from mono-cultures of cariogenic species providing an acidic challenge to enamel, to the development of diverse mixed cultures growing on a habitat-simulating medium in an artificial mouth or chemostat. The latter systems are of value in determining either mechanisms of action or cause-and-effect relationships--e.g., between dietary components or antimicrobial agents and the microflora. Laboratory models have also shown that the sensitivity of oral bacteria to inhibitors is markedly reduced when growing in biofilms such as dental plaque. Animal models have proved unequivocally that caries is an infectious and transmissible disease. Their use has enabled comparisons to be made of (a) the cariogenic potential of different bacterial species, (b) the role of the diet, and (c) the effects of potential anti-caries agents. It has been claimed that no caries-protective agent currently in use has failed a rodent test. In situ human models have been designed to permit the development of "natural" plaque on standardized enamel surfaces freely exposed to the human oral environment. The microflora that develops on unadulterated surfaces is similar in composition to that found at comparable sites on vital teeth. Demineralization can be accelerated by the inoculation of additional cariogenic bacteria coupled with either intra- or extra-oral sucrose rinses. The increased realism associated with the transition from laboratory to human in situ models is countered by a reduced ability to control or manipulate the system for experimental purposes. Thus a hierarchy of tests is needed for the study of anti-caries agents, each requiring a varying degree of microbiological complexity.