Drugs of abuse produce amnestic effects in humans and laboratory animals in a variety of tasks. Generally, only a few compounds have been examined in any particular procedure. It was the goal of the present studies to examine drugs of abuse of different pharmacological classes in rats responding under two behavioral schedules historically employed as experimental models of memory: spatial alternation and matching to position. One group of rats responded under a single-response spatial-alternation baseline with a 10-s delay and another group responded under a matching-to-position baseline with delay values of 3, 10 and 30 s. Performance under the spatial-alternation baseline was characterized by low variability and >90% accuracy. Under the matching-to-position baseline, saline control percent accuracy was >95% at 3 s, >85% at 10 s and >70% at 30 s. Under spatial alternation cocaine, d-amphetamine, pentobarbital, diazepam, phencyclidine, scopolamine and methscopolamine produced significant (P<0.05) effects on accuracy, whereas only cocaine, d-amphetamine, pentobarbital and phencyclidine disrupted accuracy under the matching-to-position baseline. These results suggest that spatial alternation may be a more sensitive baseline for determining drug effects on working memory in the rat.