In this paper we examine the impact of amnesia and of levels of processing on implicit memory by using a well-established fragmented picture-naming test with proven adequate reliability. A group of patients with amnesia of non-Korsakoff etiology was compared to a control group. While amnesic patients showed a deficit in perceptual priming, both groups showed a comparable level of processing effect. Our results confirm that when a reliable implicit memory test is used amnesia and levels of processing can both be shown to affect implicit memory performance. Thus, functional dissociations between explicit and implicit memory tests may be the consequence of a methodological artifact, that is, a difference in the reliability of the tests.