A wide diversity of animal models has been used to examine antidepressant activity. These range from relatively simple models sensitive to acute treatment, to highly sophisticated models that reputedly model some aspect of depressive illness and which yield a positive response to prolonged, chronic, drug treatment. In recent years antidepressant drug research has focused on the search for antidepressant therapy that has a more rapid onset of action. To be relevant, therefore, animal models must measure the time course of drug action. This review examines the claims of animal models to be sensitive to chronic drug treatment and considers their relevance. First, the review addresses the criteria necessary to examine the validity of animal models of depressive illness. Second, those animal models sensitive to chronic antidepressant treatment are reviewed with respect to their validity as animal models of either depressive illness and / or antidepressant activity. In particular, the development and utility of two ethologically-relevant animal models, the resident-intruder and social hierarchy paradigms, are described in detail. These models of rodent social and agonistic behaviour demonstrate that acute and chronic treatment with antidepressant drugs (regardless of their acute pharmacological activity) induce diametrically opposite changes in rodent agonistic behaviour. It is argued that the common ability of chronic treatment to increase rodent aggression (which in turn results in increased hierarchical status in closed social groups) most likely reflects the increased assertiveness and associated externalization of emotions expressed during recovery from depressive illness. Finally, findings that relate observed behavioural changes to underlying neurochemical changes are briefly reviewed.