Square mesh panels are often inserted in the codend or extension of a fishi
ng net to allow under-sized fish an extra opportunity to escape. The effect
iveness of the panel can be quantified by placing a small-mash cover over t
he panel to catch all the fish that escape through it. This paper develops
methodology for modelling the data From such covered panel experiments. The
selection process is modelled in two stages: first. selection in the panel
area, and second. conventional codend selection. Several different curves
describing selection in the panel area are investigated. These include curv
es that separate the process of fish "contacting" the panel from the select
ion of the panel itself. Parameter estimation. inference, and methods for a
ssessing goodness-of-fit are described. The methodology is illustrated usin
g haddock and whiting data.