Previous research has demonstrated a phenomenon known as the misinformation effect: After viewing an event, witnesses given misleading information are more likely than controls to report details consistent with that information. In Study 1, witnesses watched a video depicting a car chase and then completed a questionnaire containing either misleading or accurate information. Later we tested their memory for the video. Among nonmisled witnesses, high scores on self-deception and impression management measures correlated positively with accuracy. No correlation was found among misled witnesses. With additional evidence from Study 2, we argue that the results support the memory impairment explanation for the misinformation effect and do not support any of the other 3 explanations researchers have proposed.