The search for a better understanding of cognitive decline in man has lead to the use of increasingly complex procedures in animal research. The analysis of the data generated in such experiments has been greatly facilitated by the wider use of computer assisted techniques. These techniques can only be as good as the hypotheses they are used to test. Signal detection theory (SDT) provides a rational framework within which to work. The procedures are derived from human cognitive neuropsychology and are already used to some extent in primate but to a lesser degree in rodent research. The use of SDT offers two main advantages: first, a testable hypothesis as to the manner in which competing processes arrive at choice between various courses of action; second, the statistical procedures offer clear advantages over more traditional approaches by reducing the chances of misinterpretation. Though relatively easy to apply some care must be exercised in the protocol design and the choice of SDT indices if the full value of the approach is to be achieved. If experimental designs can be developed to include the appropriate use of SDT analysis; both the power of such protocols, and their value in the understanding of cognitive function, will provide a major step forward for animal-based research.