Audio playback of vocalizations by conspecifics is commonly used to elicit calls when surveying birds of prey. Methods for call surveys vary widely in their use of silent listening periods, and usually range from 3–15 min in length. We aimed to refine this approach for detecting Great Horned Owls (Bubo virginianus) in Arctic Alaska, which is the northernmost limit of their breeding range. We used two playback protocols: protocol 1 entailed uninterrupted playback, whereas protocol 2 interspersed silent listening periods with playback during 12-min surveys. In playback surveys consisting of 166 point counts during the 2017 and 2018 breeding seasons, the probability of detecting a Great Horned Owl was 0.46 (95% CI ¼ 60.09) with protocol 1 and 0.35 (95% CI ¼ 60.12) with protocol 2 (P ¼ 0.18). The probability of detection rose with the length of the playback: of all owls detected during the 12-min surveys, 23% (95% CI ¼ 66.4%) responded within the first 3 min, 52 6 7.6% within the first 6 min, and 80 6 6.1% within 9 min. Including silent listening periods was not necessary for detecting Great Horned Owls during call surveys. We found no correlation between probability of detection and either cloud cover or wind speed (P ¼ 0.60 and P ¼ 0.28, respectively). However, we found a negative correlation between temperature and probability of detection (P ¼ 0.02). From these surveys, we calculated the density of Great Horned Owls in the Middle Fork Koyukuk Valley, Alaska (approximately 67.589°N, 149.789°W) was 4.2 6 2.6 owls/km² during the winters of 2017 and 2018, which represents the first estimate of density at the northern breeding limit of the species.