Are amnesic patients selectively impaired in recall relative to recognition? Experiment 1 studied a group of Korsakoff amnesics and matched the amnesic level of recognition with that of control subjects by testing control recognition of unrelated words after longer delays. It was found that under these conditions the observed levels of recall were also approximately equal. In Experiment 2, a similar result occurred when the Korsakoff amnesic level of recognition for unrelated words was matched by varying the number of presentations as well as delay before testing. In Experiment 3, a similar result occurred again with a group of amnesics of mixed aetiology and recognition levels for related words matched by varying the duration of presentation and delay before testing. In this experiment, both recognition and recall of the same items were assessed. It was found that for the amnesic group the observed level of dependency between recognition and recall was less than that for the control group. One possible interpretation of this result is that the microstructure of the recall process may be selectively disturbed in amnesia. The principal finding, however, is that in all three experiments there was no significant evidence of the existence in amnesia of a selective deficit in the overall level of recall relative to that of recognition.