Selection by the codend has been the subject of much research over the past thirty years, with trials in artisanal and industrial fisheries around the world. These studies typically test only a few gears, partly for logistic and/or economic reasons and partly to ensure there are sufficient hauls to estimate the selection of each gear with reasonable precision. To explore a broad range of selective gear options for use in a fishery, and to understand better the relative influence of the important variables related to gear design, it is necessary to develop models that predict selection across all of these variables. Such empirical models are best constructed in meta-analyses that combine the data from many trials. There are, however, few meta-analyses in the size-selection literature and these usually only consider the effect of codend mesh size. In order to make best use of existing selectivity data we review the meta-analyses that have taken place and extend these studies to investigate the vertical distribution of fish at the mouth of trawl to estimate selectivity for data poor species.These analyses advance our understanding of the potential of using selective gears to reduce discards and will be of particular importance in relation to potential ‘choke species’ which may not have had much attention in the past.