Research on species selection is one of important potential measure to alleviate the bycatch and discard problem of trawl fisheries. More and more research showed that parametric models, which have been widely used in analysis of selective devices of trawls, were not only important to assess and compare the performance but indicative to improve the construction of the selective devices. In this paper, a model for assessing the separating efficiency of separator devices rigged in multilevel beam trawl fishing offshore of the East China Sea, in which the size selection of separator panel was taken into account and encounter probability (presented by ogive, Logistic curve) of individual entering trawl was regarded as dependent upon the individual size and then presented by ogive (Logistic curve), was developed based on the principle of Select model. Fitness of model for size frequency of Parapenaeposis hardwickii and Cynoglossus abbreviatus caught by multilevel beam trawl rigging separator panel with 60 mm mesh size and the maximum likelihood estimates of parameters showed the model was applicable to assess the separating efficiency of panel. According to the discussion, the model could be applied to the other selective devices and simplified by hypothesis test method according to different situation. Additionally, we think the encounter probability of individual, which was independent of size selection of selective device, could be used to analyse the responding behavior of aquatic animal.