We have shown earlier that mild electric shocks induce a lasting social avoidance in male rats. Here we investigated whether shock-induced social avoidance can be developed into a laboratory model of stress-induced anxiety. The putative new model would assess sub-chronic, stress-induced anxiety (as opposed to tests based on natural fear) in a heterologous context (as opposed to classical fear conditioning). A single exposure to mild electric shocks induced a robust social avoidance that lasted more than 5 days. Low doses of chlordiazepoxide (0.5, 1 mg/kg), diazepam (0.5, 1, 5 mg/kg), buspirone (0.3, 1 mg/kg), and fluoxetine (1, 3, 5 mg/kg) abolished this effect, whereas the anxiogenic compound m-chlorophenylpiperazine (0.5-3 mg/kg) induced social avoidance in unshocked rats. These effects were produced at doses that did not affect locomotion in the open field. Haloperidol (0.05, 0.1, 1, 5 mg/kg) influenced social avoidance at sedative doses only. The sensitivity of the model to anxiolytic agents was compromised at high (sedating) doses. Taken conjointly, these data show that shock-induced social avoidance can be used to assess the anxiolytic potential of compounds. In addition to predictive validity, the model appears to show construct and face validity as well: stress is among the etiological factors of, whereas social avoidance simulates the social deficits seen in, a variety of anxiety disorders. The model may be used to study the effects of anxiolytics on sub-chronic states of stress-induced anxiety.