A new method involving the blockade of operant behaviour induced by the withdrawal of a conditioned signal for safety without presentation of a punishment signal has been developed for studying drugs with anxiolytic or anxiogenic properties. For this purpose, rats were trained under two alternating components of a multiple schedule of reinforcement FR8 (food)/FR1 (food) + RR 50% (shocks randomly delivered with 50 +/- 15% of the presses). The nonpunished and punished periods were signalled by one cue light above the right lever (safety signal) or the left lever (punishment signal), respectively. On the test session (safety signal withdrawal), the safety signal was turned off at the end of the first nonpunished period, but the punishment signal was not presented (every press was food rewarded and no shocks were delivered). During this period (4 min), rats exhibited a strong blockade of responding that lessened over time. This suppression seemed not to be caused by intervening events such as novelty, temporal conditioning, schedule of food delivery or ambiguity of the signal presented. The behavioural blockade induced by withdrawal of the safety signal was reduced by benzodiazepines: diazepam (0.5-4 mg/kg), chlordiazepoxide (4-8 mg/kg), nitrazepam (0.25-2 mg/kg), alprazolam (0.25-1 mg/kg), and partial agonists at benzodiazepine receptors: bretazenil (0.125-8 mg/kg) and ZK 91296 (32-64 mg/kg). Various 5-HT-related drugs also lessened the behavioural blockade:pCPA (3 x 150 mg/kg) and the 5-HT1A receptor agonists, buspirone (0.25-2 mg/kg), gepirone (0.25-1 mg/kg) but not 8-OH-DPAT. Compounds that may cause anxiety in humans further enhanced the blockade of lever pressing induced by the safety signal withdrawal at doses that did not modify baseline responding: d-amphetamine (0.125-0.5 mg/kg), caffeine (16 mg/kg) and picrotoxin (1 mg/kg). FG 7142 (8 mg/kg) and CGS 8216 (2-8 mg/kg) decreased responding during both components of the session. Therefore, the present paradigm seems sensitive to both "anxiolytic" and "anxiogenic" effects of drugs under identical procedural conditions.