By using in situ models, we have the potential to study both fundamental aspects of the caries process as well as more applied research problems such as the effect of food on dental caries and the role of fluoride in caries prevention in human subjects without actually causing caries in the natural dentition. The key experimental parameters that need to be considered in the development of an in situ model are the characteristics of the subject panel, the physical design of the model, the type of hard tissue substrate and the method of assessing mineral status, and the study design and clinical protocol. Each parameter must be carefully considered in relation to the objectives of the research, study design requirements, ethical considerations, impact on clinical relevance, and impact on the control of variation. The major source of variation associated with in situ models should be of biological and not experimental origin. The design and conduct of proper in situ model studies require a clear understanding of the caries process, sound analytical support, and a knowledge of how to work with research subjects to achieve a high level of compliance. Given the complex nature of caries, a combination of hard tissue substrates--including sound, surface-softened lesions and subsurface lesions--may be necessary to model all aspects of caries progression and prevention successfully. Internal validation of in situ models using fluoride dose-response controls is considered to be necessary for studies evaluating the efficacy of new fluoride dentifrice formulations.