The cover net method is widely used in testing fishing gear selectivity. However, there exists influence of this method on the mesh selectivity of codend because of the blocking ( mask ) effect . In this paper, the selective characteristics of the codend by cover net method were compared with those by the parallel haul method based on the square mesh codends with the mesh sizes of 25, 30, 35, 40 and 45mm respectively. The selectivity curves were described by the logistic equation. The selective parameters and the main selective indexes were estimated by the maximum likelihood method in the case of cover net method, and by the SELECT model when using parallel haul method. The standard errors of the parameters were estimated by the Fisher information matrix and the goodness-of-fit of dif ferent models was assessed by the AIC ( AKAIKE information criterion) . The selectivity of different codends was obtained by the analysis of the catch data of anchovy ( Setipinna taty ) . In the selective analysis of parallel haul method, it was found that the equal split model was f itted better than the estimated split model through the comparison of the AIC. Then the main selective indexes est imated from different experimental methods were compared and the result demonstrated that the 50% retention lengths, the selective ranges and the selective factors obtained from the parallel haul method are respectively 7. 8 percent larger, 24. 5 percent less and 7. 4 percent larger than those estimated from the cover net method, but no signif icant differences ( P> 0. 05) among them were found. If the minimal mesh size of stow nets is determined based on the result from the cover net method, the mesh size will tend to be bigger, resulting in excessive escape of the fish inside the codend.