Fishing experiments were carried out in Yeongil Bay, Korea using an encircling gill net with four different mesh sizes and two different hanging ratios to measure the mesh selectivity for gizzard shad Konosirus punctatus. Twenty-six trials were conducted giving a total catch of 485 gizzard shad. The 'share each length's catch total' (SELECT) analysis with maximum likelihood method was used to fit the different functional models, the normal, lognormal, and bi-normal model, for a selectivity curve to the catch data. In addition, two cases in which the relative fishing intensity was either estimated or fixed were compared. The bi-normal model with the fixed relative fishing intensity was found to fit the data best. For the selectivity curve on the bi-normal model, the relative length (the ratio of fish length to mesh size) with the maximum efficiency was obtained as 3.70. From this, the focal fish lengths in the commercial encircling gill net, with mesh size ranging from 5.0 to 6.0 cm, were inferred to be 18.5-22.2 cm.KEY WORDS: bi-normal model, encircling gill net, gizzard shad, maximum likelihood method, mesh selectivity, SELECT.