Summmy.-Laboratory techniques for assessing the behavioral effects of anxiolytic drugs often involve the use of a punishing procedure and rats or other species as subjects. Some aspects of conventional experimental designs, however, make difficult interpretation of the drugs' effects. The present experiment made use of a modified punishment with rats, which was designed to reduce interpretative difficulties. Both food and shock presentations were scheduled at similar, though indepeqdent, variable intervals. Chlordiazepoxide and amobarbital were found to produce dose-related decreases in the suppression of responding produced by punishment.Techniques of punishment have been widely used for investigating the effects of psychoactive drugs and have proved useful for predicting the clinical potency of potential anxiolytics (Cook & Davidson, 1973;Houser, 1978). The most widely studied of such procedures was first described by Geller and Seifter ( 1960). Rats are trained to press a lever to obtain food or water reinforcement which is made available according to a variable-interval schedule. Then a signal, which often lasts for 3 min., is presented at regular intervals, e.g., once each 15 min. During these signalled periods, the schedule is changed so that each response now produces a reinforcer but also produces an electric shock. Relatively low rates cif responding are usually observed during the signal, the actual rate being dependent upon the parameters of the shock used (Geller, e;al., 1962). A number of experimenters have shown that benzodiazepines, barbiturates, and other drugs which have been used clinically as anxiolytics can increase the low rates of punished responding maintained by this procedure (Blum, 1970; Geller, et al., 1962).This facilitation of punished responding has proved to be one of the most replicable phenomena in behavioral pharmacology. There are, however, several aspects of the Geller-Seifter procedure which make such results difficult to interpret. ( a ) The schedule of food presentation is changed during the periods of punished responding, and so comparisons becween punished and unpunished levels of responding are not possible within any session. (b) The use of a continuous reinforcement schedule of food and of shock delivery involves a direct association between food or water reinforcement and shock, (c) Any