Animal disease models are considered important in the development of drugs for Alzheimer's disease. This brief review will discuss possible reasons why their success in identifying efficacious treatments has been limited, and will provide some thoughts on the role of animal experimentation in drug development. Specifically, none of the current models of Alzheimer's disease have either construct or predictive validity, and no model probably ever will. Clearly, specific animal experiments contribute to our understanding of the disease and generate hypotheses. Ultimately, however, the hypothesis can only be tested in human patients and only with the proper tools. These tools are a pharmacologically active intervention (in humans) and a clinical trial suited to evaluate the mechanism of action. Integration of knowledge in quantitative (sub) models is considered important if not essential in this process.