The present study aimed to index the accuracy of memory for acute trauma symptoms by comparing the symptoms reported by motor vehicle accident (MVA) victims within 1 month posttrauma with the recall of these symptoms at 2 years posttrauma. Ninety-two consecutive MVA admissions were assessed for the presence of acute stress disorder (ASD) within 1 month posttrauma. At 2 years posttrauma, 61% (N = 56) of the sample were reassessed for posttraumatic stress disorder (PTSD) and for accuracy of recall of the symptoms reported during the first assessment. At least one of the four ASD diagnostic clusters was recalled inaccurately by 75% of patients. High levels of posttraumatic stress severity and high subjective ratings of injury severity at 2 years posttrauma were associated with errors of addition (i.e., recalling the presence of acute symptoms 2 years posttrauma that were not reported during the first assessment). Low levels of posttraumatic stress severity and low subjective ratings of injury severity at 2 years posttrauma were associated with errors of omission (i.e., omitting to recall acute symptoms 2 years posttrauma that were reported during the first assessment). These results suggest that retrospective reports of acute stress symptoms should be interpreted cautiously because of the influence of current symptoms on recall of acute symptoms.