The aim of this paper is to examine from a statistical viewpoint the model and the estimation performance for covered-codend experiments in which fish escape from the covernet. The logistic function was assumed for the selectivity curves of the codend and covernet. First, we evaluated the efficiency of the maximum likelihood (ML) estimation based on the model that considers fish escape from the covernet compared with the conventional model that does not consider the effect of fish escape. Second, we assessed the biases of the ML estimators of parameters in the codend selectivity curve by using the conventional model when fish escape occurred. Finally, we examined the performance of model selection based on the Akaike information criterion through Monte Carlo simulations. It is concluded that the model considering fish escape is effective for the analysis of selectivity. In addition, the estimator involving the model selection performed better than estimators based on fixed models. The interpretation of these results and topics of future research are also discussed.