The accuracy of children's long-term memory for distressing events is of theoretical and legal interest. In this longitudinal study, 3- to 5-year-olds and their main parental caretakers individually participated in a mildly distressing event, and 1 h later the children's memory/suggestibility was assessed (Time 1). Six to seven years later (Time 2), the children (9- to 12-year-olds, N = 54) were again interviewed about the Time 1 experience. Their memory was more accurate at Time 1 than Time 2. Children who were rated as more distressed at Time 1 recalled more units of correct information at Time 2 if the children were older or lower in self-reported attachment anxiety. Children who were older at Time 2 were less suggestible. Time 1 misinformation did not contaminate Time 2 free recall. Theoretical and applied implications are discussed.