In Japan, wildlife managers are considering night culling as a new tool for managing sika deer populations, but there is no information on effective night culling practices. We investigated the relationships among distance from the observation sites, luminous intensity, and deer behavior on Nakanoshima Island, Hokkaido, Japan, from October 18 to November 19, 2016. We used two kinds of lights and recorded the luminous intensity, number of deer, herd size, and flight behavior, including the time to initiation of flight. In addition, we used infrared-triggered cameras to clarify the influence of the distance between the observation sites and the bait sites on deer appearance frequency. We found a significant relationship between flight frequency and herd size under conditions of high luminous intensity. The proportion of flight behavior decreased as distance from the observation sites increased. In addition, the proportion of flight behavior under high luminous intensity tended to be higher than that under low luminous intensity. Moreover, more deer were photographed beyond survey time than during survey time at bait sites that were 25m and 50m away from the observation sites. We suggest, therefore, that setting bait sites at a distance of 50-100m from the observation sites would likely cause the following outcomes: deer would appear more frequently at bait sites, deer would be less likely to flee, and the success of culling would increase. Wildlife managers would have to use adequate lighting to maintain sufficient visibility and safety without heightening the alert responses of the deer.