This paper examines the role of orthographic-to-phonological mapping in memory. Specifically, we consider the hypothesis that distinctive orthographic-to-phonological mapping (Hirshman and Jackson, in press) mediates the testing effect. We test this hypothesis by examining whether manipulations designed to influence the distinctiveness of orthographic-to-phonological mapping interact with the testing effect. Consistent with our hypothesis, the testing effect is minimized under conditions promoting distinctive orthographic-to-phonological mapping. A testing effect occurs for regular words presented in a nonsense word context or in isolation, but not for regular words presented in an exception word context or for exception words presented in a nonsense word context. Further, a control experiment using a levels-of-processing manipulation demonstrates that these effects are specific to testing. We close by presenting a conception of how orthographic-to-phonological mapping representations interact with other memory representations during retrieval.