The long-term stability of youth reports of traumatic events is largely unknown. Translational animal research suggests that there may be an alteration of memories for traumatic events via memory reconsolidation processes, whereas clinical research suggests memory alteration may occur through augmentation by negative emotions. In this report, 2 natural experiments test reconsolidation model and augmentation model predictions about the course of traumatic memories in youth. Data are from 2 prospective studies that assessed reports of an initial traumatic event (Hurricane Katrina) and tested recall both pre and post a similar event (Hurricane Gustav). In the 1st (Sample 1; n = 94, initial Grade 9 followed to 11), youth were assessed at 4 time points: Times 1-3 were 13, 20, and 26 months post-Katrina and then Time 4 was 5 months post-Hurricane Gustav. In the 2nd (Sample 2; n = 141, Grades 4 through 8), youth were assessed at 12 months pre-Gustav (Time 1; 24 months post-Katrina) and then again at 1 month (Time 2) and 8 months (Time 3) post-Gustav. Those with relatively high Gustav exposure showed more stability in their reports of Katrina exposure events, whereas in those with low Gustav exposure, reports of Katrina events decreased. Time spans between recall, age, gender, symptoms of posttraumatic stress disorder, or cognitive/learning ability did not explain changes in the reports. The study provides the 1st long-term data on the consistency of youth reports of disaster-related experiences and provides initial evidence for the ecological validity of memory reconsolidation theory applied to traumatic events in youth.