Poor selectivity of demersal trawls targeting fish, crustaceans and cephalopods in Mersin Bay, Turkey, (and more broadly for most Mediterranean demersal trawlers) is a significant concern. The majority of the boats working in the Mersin Bay trawl fishery use traditional gears with 44mm mesh size hand-woven slack knotted codends in which approximately 50% of the catch by weight and 70% by numbers are discarded. The present study compares size selectivity of a commonly used hand-woven codend and three alternative machine woven codends (40 mm square mesh; 44 mm and 50 mm diamond mesh) for seven commercial species: red mullet (Mullus barbatus); brushtooth lizardfish (Saurida undosquamis); common pandora (Pagellus erythrinus); goldband goatfish (Upeneus moluccensis); Randall's threadfln bream (Nemipterus randalli); green tiger prawn (Penaeus semisulcatus); and speckled shrimp (Metapenaeus monoceros). A total of 87 hauls were conducted using a covered codend method in 2011. Short term commercial loss for 31 marketed species resulting from switching from the commercial codend to each alternative codend, was estimated. Results show that size selectivity of the commercial codend is rather poor for almost all the marketed species. The 40 mm square mesh codend is the best alternative for the majority of the marketed species in terms of releasing juveniles. We estimate a commercial loss of 17% in landing values if this codend is used. (C) 2014 Elsevier B.V. All rights reserved.