KEY woms-Animal models, research, depression, old age.Animal models of depression are often thought to be of little relevance for understanding the clinical disorder, since the behaviours involved frequently carry little conviction as analogues of human behaviour. However, to dismiss animal models on these grounds is to misunderstand their purpose-or rather purposes. The most familiar usage of animal models of depression is within the pharmaceutical industry, as screening tests in the development of novel antidepressants. As the only purpose of drug screening is to produce new drugs, the extent to which the procedures used resemble the clinical disorder is largely irrelevant: the major issue is whether a test successfully predicts clinical efficacy. This article is concerned with a different aspect of animal models of depression, their use as simulations for investigating the psychobiology of depression. Many of the available models were developed primarily as predictive drug screening tests, without regard to their validity as simulations of depression. However, a minority of models have been developed explicitly for this purpose; and less valid models can also provide useful corroborative insights.The procedures for validating animal models of psychiatric disorders have been discussed in detail elsewhere (Willner, 1984(Willner, , 1986(Willner, , 1991; they include consideration of predictive validity (which concerns primarily the correspondence between drug actions in the model and in the clinic), face validity (phenomenological similarities between the model and the disorder), and construct validity (a sound theoretical rationale). Some desirable features in a simulation of depression are that the model should respond to antidepressant drugs; should employ realistic inducing conditions; and should model a core symptom of the disorder. In addition, for many purposes, such as investigating mechanisms of antidepressant action over a clinically relevant time scale, a prolonged time course is also desirable.