Rose shrimp (Parapenaeus longirostris) and Norway lobster (Nephrops norvegicus) are the two main target crustaceans for the demersal trawl fleet in the international waters of the Aegean Sea. In the present study, selectivity data were obtained for these two species in commercial codends (300 MC is 40 mm polyethylene codend with 300 meshes on its circumference) and two newly designed types, known as a narrow codend (200 MC is 40 mm polyethylene codend with 200 meshes on its circumference - 33% reduced) and square mesh top panel codend (SMTPC is constructed as 150 diamond meshes on the lower and 75 square meshes on the top panel), respectively. These three types of codends were tested using the covered codend method. Trawling was carried out at depths ranging from 269 to 426 m onboard a commercial vessel in August 2004. Selection parameters were obtained by fitting a logistic equation using the maximum likelihood method. The rose shrimp, in terms of weight, was the most abundant marketable species, with catch per hour values of 19.1, 16.0, and 15.0 kg in 300 MC, 200 MC, and SMTPC, respectively. Results of the selectivity analysis indicate that the commercially used 40 mm nominal mesh size PE codend is rather unselective for these species. In general, narrow and square mesh top panel codends have relatively higher L(50) values than the commercial codend. However, the differences are significant only for rose shrimp, but not for Norway lobster. It is concluded that despite the improvements in selectivity, the codend modifications 'narrow' and 'square mesh top panel' were not sufficient to release immature specimens of either species, at least not when using the 40 mm PE standard netting.