An experimental study of extinction of conditioned drug-induced effects was carried out to determine: 1) duration of the extinction effect; and 2) stability of extinction as determined by a challenge with a stimulant drug. Twelve animals with unilateral 6-hydroxydopamine (6-OHDA) substantia nigra lesions were assigned to paired and unpaired treatment groups (n = 6) in a Pavlovian conditioning paradigm. The paired animals received apomorphine (0.05 mg/kg SC) immediately prior to placement into a test chamber and the unpaired animals received the apomorphine 30 min following test chamber placement. The two groups were matched for apomorphine-induced contralateral rotation prior to the conditioning treatment. Following Pavlovian conditioning, the paired group, but not the unpaired group, exhibited contralateral rotation in a nondrug test trial. This conditioned response underwent extinction after one nondrug extinction trial and the extinction effect persisted for 2 months. When tested with caffeine (10 mg/kg), the paired animals again exhibited substantial contralateral rotation. In contrast, the unpaired animals showed only an increase in ipsilateral rotation in response to the caffeine treatment. The drastically different response to caffeine in the paired and unpaired animals was not due to prior apomorphine exposure per se or due to 6-OHDA lesion-induced differences in striatal dopamine depletion. Rather, the effect of caffeine on rotation behavior was determined by the Pavlovian drug conditioning procedures carried out several months earlier prior to caffeine testing.