After people are given new and misleading information about a previously seen incident, they are often unable to remember the original incident accurately. Loftus (1979;Loftus & Loftus, 1980) argued that the forgetting observed under these conditions is due to the loss of original memories that occurs once new information has been encoded. However, closer inspection of the testing procedures used in the Loftus studies suggests an alternative explanation. It was hypothesized that the retrieval environment will lack sufficient cues to ensure the access of original memories when test materials are randomized with respect to the original sequence. To investigate this possibility, one half of the subjects received test slides in a random order, as reported in Loftus, Miller, and Burns (1978, Experiment 1). The other half received slides in an order that matched the order seen during input. Forgetting original memories occurred only under the random test-order condition. Thus, there was no evidence of loss of original memories. Rather, it appears that forgetting produced under these procedures is due to the absence of critical cues at the time of retrieval that were present during original encoding.Both authors contributed equally to the ideas expressed in this article. We wish to thank M. Robinson for the recruitment and running of all subjects. We also would like to thank John Morton for helpful comments.Requests for reprints should be sent to D.

This article reports on research into the issue of exactly when false information biases memory for a witnessed event. A total of 124 college students were first shown a series of slides depicting an autopedestrian accident. They were then given either inconsistent or consistent postevent information on a written questionnaire. The order of questionnaire items was either sequential or random, that is, either matched or did not match that seen originally. Finally, a recognition test for slides was administered, with the test items either m a sequential or random order Original memory was biased by postevent information except under a random-questionnaire, sequential-test condition. These results are compatible with a retrieval-based explanation of forgetting Implications for interrogation procedures used in legal settings are considered m the light of these and other results.