Chronically administered antidepressant drugs, particularly selective serotonin (5-HT) reuptake inhibitors (SSRIs), are clinically effective in the treatment of all anxiety disorders, while the clinical effectiveness of "traditional" anxiolytics, such as benzodiazepines (BDZs), is limited to generalised anxiety disorder or acute panic attacks. This implies that animal models of anxiety should be sensitive to SSRIs and other antidepressants in order to have predictive validity. We reviewed the literature on the effects of antidepressants in the so-called animal models of anxiety and found that only the isolation-induced calls in guinea-pig pups may reveal anxiolytic-like action of all antidepressant classes after acute administration. Some other models, such as marble-burying or conditioned-freezing behaviours, and isolation- or shock-induced ultrasonic vocalisation models, may detect anxiolytic-like activity of acutely administered antidepressants, although the sensitivity of these models is usually limited to SSRIs and other drugs affecting 5-HT uptake. The predictive validity of models of "anxiety", such as the plus-maze and light-dark transition tests or stress-induced hyperthermia, appears to be limited to BDZ-related drugs. Far less work has been done on chronic administration of antidepressants in animal anxiety models. Unless and until such studies have been undertaken, the true predictive value of the anxiety models will remain unknown.