A danger response evaluation device estimates the content of a possible danger through simulation using observation data acquired from an external device. Then, when the estimated content of the danger exceeds a predetermined threshold value, the danger response evaluation device gives, to a user terminal of a user, notification of an alarm that prompts a response to the danger. Thereafter, when having acquired the content of a response of the user to the danger from the user terminal of the user, the danger response evaluation device determines how much the content of the response corresponds to execution of a plan of a response to the danger stored in a storage unit in advance. Then, the danger response evaluation device evaluates, based on a result of the determination, how much the content of the actual response of the user contributes to a damage prevention effect related to the danger.