Parallel gillnet experiments were carried out in order to estimate catch efficiency of gillnet and abundance of chum salmon in the central Bering Sea in July, 1984. In the experiment, two gillnets were simultaneously set in parallel configuration at a distance of about 2km, and the catch in number of salmon by each side of each gillnet was recorded. Release and recovery ex-periments were also conducted to estimate the moving speed of salmon. In order to estimate catch efficiency and density of fish, data were analyzed based on a model that takes into account the moving speed of fish and the catch interaction between two parallel gillnets. A relationship between CPUE (number/tan) and density (number/km²) was constructed, and CPUE at each station was converted into density. The density was weighted by the size of the survey area, and the abundance of chum salmon was calculated.