The primary objective of this study was to determine whether brief exposure to a stressor would provoke relapse to cocaine-trained behaviour and, if so, whether such an effect could be blocked pharmacologically. Rats were initially trained to self-administer cocaine HCl (1.0 mg/kg/infusion, IV; one 3-hour session/day; 9-12 days). Subsequently, extinction conditions were introduced in which lever-pressing resulted in IV infusions of saline rather than of cocaine. Extinction conditions were maintained until animals made 15 or fewer responses in the 3-hour session, after which they received saline infusions at the start of each session until they made 10 or fewer responses in 3 hours. Subsequently, animals were tested for reinstatement of responding for saline infusions following a non-contingent injection of cocaine (2.0 mg/kg, IV) and exposure to intermittent footshock (10 min, 0.5 mA, 0.5 sec on, mean off period of 40 sec). In Experiment 1, footshock stress induced reinstatement of cocaine-trained behaviour after prolonged extinction and after a 4- to 6-week drug-free period; an effect comparable to that induced by a priming injection of cocaine. In Experiment 2, animals were given saline, cocaine, and footshock tests for reinstatement 20 min after pretreatment with saline and diazepam (0.75 or 1.5 mg/kg, IP). In this experiment, diazepam attenuated footshock- but not cocaine-induced reinstatement of cocaine-trained behaviour. These findings suggest that the neurochemical events mediating footshock- and cocaine-induced reinstatement of cocaine-trained behaviour are pharmacologically dissociable.