This study investigates perceptions of and memory for a filmed ambiguous event, intended to simulate features of a contentious naval incident that occurred during World War II. Participants viewed a short film that contained elements attributable to a storm or a battle at sea. In different conditions, test instructions mentioned speculation about the possibility of a storm or a battle, or were neutral. Participants exposed to the battle prime were significantly more likely to describe a battle taking place than were participants exposed to either a storm or neutral prime. Evidence of the influence of expectations was also obtained via a recognition measure and confidence ratings. Memory biases were unchanged 7 weeks post the initial viewing. It is concluded that observers of ambiguous events during times of war are vulnerable to errors based on schematic expectations and that these patterns of errors can be replicated in laboratory simulations.