<bold>Abstract</bold>A recent study revealed that fat greenling <italic>Hexagrammos otakii</italic> produces call sounds during the spawning season; the number of fish calls increases on the spawning day. However, it is unknown where and how many individual fat greenling produce fish calls. This
behavior of fat greenling could be estimated by localizing the fish calls. We propose a sound localization method that uses two stereo recorders, which can then be used to estimate two time differences of arrival for each fish call. First, the calls of fat greenling are recorded by two stereo
recorders for 3 weeks during the spawning season. Second, the calls of fat greenling are detected from the recorded data by the automatic detection algorithm, which uses the acoustic features of the fish call, and the positions of fish calls are estimated by the proposed sound localization
method. The analysis of the recorded data by using the proposed method localized most of the fish calls near the spawning bed. Furthermore, the number of calling individuals may be estimated by analyzing the relation between the time the fish call is produced and its localized position.