Three groups of amygdala-kindled rats received 10 bidaily treatment trials: On each trial, the drug-before group received a diazepam (2.5 mg/kg i.p.) injection 1 hr before a convulsive stimulation, the drug-after group received a diazepam injection 1 hr after a stimulation, and the vehicle control group received a vehicle injection either 1 hr before or 1 hr after a stimulation. After treatment, only the drug-before group displayed significantly longer forelimb clonus under the influence of diazepam (that is, they displayed contingent tolerance to diazepam's anticonvulsant effect) and significantly longer forelimb clonus while drug free. Following a 14-day retention period, the rats in the drug-before group retained significant levels of contingent tolerance but did not display significant increases when tested drug free. These data suggest that compensatory responses do not play a causal role in the expression of contingent tolerance.