Although there is considerable agreement that performance in direct memory tasks (e.g., recall, recognition) is more disrupted by amnesia than performance in indirect memory tasks (e.g., mirror reading, word completion), one may be able to further circumscribe the deficit within the domain of direct memory tasks. The present article explores whether recall is disproportionately disrupted by amnesia compared to recognition. If amnesia affects memory uniformly across different direct memory measures, recall of normal controls should not differ from the recall of amnesics when recognition scores of these two groups are equated. On the other hand, if recall is disproportionately disrupted, normal recall should be superior to amnesic recall even when recognition is equated. The present study equated amnesic recognition with that of controls by providing amnesics with 8 s of study time and normal subjects with 0.5 s. Amnesics with Korsakoff's syndrome, amnesics with other etiologies, and appropriate controls were examined. Normal recall was superior to amnesic recall even when no differences were found in recognition. The results further specify the selective nature of amnesia.