Primary objective: The aim of the present study was to assess prospective memory (PM) with a video-based task using naturalistic stimuli, in a group with long-term disabilities resulting from traumatic brain injury (TBI). The specific focus of the research was on comparing the accuracy of the TBI and control groups' estimates of successful completion of the prospective memory tasks. Research design: A group of 25 persons with TBI and 20 matched controls were compared on the video-based test of prospective remembering, on a test of retrospective memory (the Logical Memory sub-test of the Wechsler Memory Scale) and tests of executive dysfunction. Correlations between the neuropsychological and PM test scores were calculated and an item analysis of the PM test was undertaken. Procedures: For the PM test, participants were given a scenario involving a robbery, and a list of tasks to be completed in a nearby city centre. The instructions were given twice and on the second occasion they were asked to rate the likelihood of remembering each item. They then watched a videotaped segment showing the perspective of a person first driving and then walking through an unfamiliar city. Their task was to recall each instruction when a relevant cue appeared. Outcomes and results: The TBI group ratings of likelihood of recalling items were entirely equivalent to those made by the controls. On the PM test, however, their performance was significantly reduced. The performance of the TBI group on the PM test was correlated with scores on the Logical Memory and verbal fluency measures. Conclusions: The results showed that, although the TBI group performed more poorly on the PM task, their expectations about how much they would remember were comparable to those of the controls. The findings suggest that persons with TBI may have unrealistic expectations about how much they will remember in a novel situation. Awareness and acceptance of cognitive changes after TBI can be a significant issue for rehabilition.