In order to find the selectivity of gillnets in the East China Sea, a production test using several gillnets of various mesh sizes (45,50 and 55 mm)and two gear heights (6 and 10 m)was carried out. Three uni-models (Normal, Lognormal and Gamma)were used to analyze the selectivity of the test nets. The parameters were estimated using maximum like lihood methods. Likelihood ratio Chi-square test and Akaike's information criterion (AIC) were used to compare the goodness of fit of the different models.Theresult showed that the fitting goodness of lognormal model is the best.The model is lower than, AIC is 147.78. The right tilt of the lognormal is obvious, the selectivity increased faster at the left side of the model peak, and gentle decline at the right side, which are well coincided with the original data.The 50% select body length of 50 mm mesh size is 123 mm, the select range is 123-209 mm. According to the population structure estimation which contacted test nets, the catch number of different body-length groups is normally distributed, the dominant range is 125-185 mm, this result is similar to previous studies. The select range of 50 mm mesh size covered the estimated dominant range, 50 mm mesh size is the most efficient for Chelidonichthys spinosus, but the influence of the mesh size on the fishery resources can not be evaluated due to the lack of knowledge about C. spinosus resources situation; the catches of the two heights test nets (6 and 10) are similar. The influence of panel heights to catches of C. spinosus is not obvious. These lesults will provide scientific guideline for coastal fishery management.