Undersized and immature commercially important cephalopods are often inadvertently caught by trawlers in the eastern Mediterranean. To evaluate the effectiveness of different mesh codends (diamond, hexagonal and square) to reduce bycatch of juvenile commercially important Mediterranean cephalopods (European squid Loligo vulgaris, broadtail shortfin squid Illex coindetii and pink cuttlefish Sepia orbignyana), a series of selectivity experiments was undertaken with a modified bottom trawl. The covered codend technique was used to capture escapees. For the broadtail shortfin squid and the pink cuttlefish, the square-shaped mesh displayed the highest 50% retention lengths (Lso) compared to diamond and hexagonal mesh. For the European squid, selectivity could only be measured by pooling the data from all hauls for each mesh shape. For all three species, the L(50) values of square and hexagonal mesh codends were significantly different (p < 0.01). Beyond the mesh variability, species catch was found to have a significant impact in the selection range of the broadtail shortfin squid. Furthermore, for the cuttlefish, total catch and haul duration likely account for variability of L(50) attributed to mesh configuration. For all species, regardless of the mesh shape, L(50) values were substantially lower than the minimum landingsize or length at first maturity. Therefore, the current legal minimum mesh size and codend configurations for demersal trawling are not suitable for the management of these species. As such, sustainability in Mediterranean cephalopod fisheries would profit from more selective gears. This could be achieved both by an increase in codend mesh size and change in codend shape; however, being part of a mixed fishery, when these changes are being practiced, the impact on the fish catches of target species will have to be taken into consideration. (C) 2009 Elsevier B.V. All rights reserved.