I examined the impact of an explicit opt-out option on eyewitness identification performance. I predicted that an opt-out option would decrease innocent-suspect identifications more than culprit identifications, and that this effect would be more pronounced when viewing conditions were worse. I randomly assigned participants (N = 2003) to watch either a clear or degraded simulated-crime video. After a brief filler task, participants viewed either a culprit-present or culprit-absent showup and responded either "Yes" or "No". Half of the participants were randomly assigned to have an additional option to respond, "Not Sure". Contrary to my prediction, the not-sure option decreased both culprit (44% to 36%) and innocent-suspect (19% to 14%) identifications; this effect was unaffected by viewing condition quality. Despite empirical evidence and theoretical rationale indicating an opt-out option would improve the culprit and innocent-suspect identification tradeoff, the present results suggest otherwise.