In an effort to improve the selectivity of beam trawls in the Belgian flatfish fishery, codend size selectivity of three flatfish and two roundfish species were compared testing 80 mm diamond-shaped mesh (T0) versus 80 mm mesh turned 90° (T90) using the covered codend method. Eight models were considered to describe the size selective performance of the T0 and T90 codends for each species; four traditional models, along with four models that considered contact probability (the fraction of fish that will not make proper contact with the codend meshes for size-dependent escape). Fifteen hauls were completed in the North Sea, fishing both codends simultaneously. The results demonstrated that the T90 codend significantly increased size selectivity for the two roundfish species investigated, whiting (Merlangius merlangus) and pouting (Trisopterus luscus), and significantly decreased size selectivity of the three flatfish species, sole (Solea solea), plaice (Pleuronectes platessa), and dab (Limanda limanda). Selection range (SR) values were similar for both roundfish species with each codend, but were significantly reduced for all flatfish species by the T90 codend. Models that considered contact probability provided the best model for four of the species with T90 (pouting, sole, plaice, and dab), and one species with T0 (sole). Lack of adequate contact with the T90 meshes could be a result of the combined effects of increased flow through the T90 codend due to T90 meshes remaining more open than T0 meshes during towing, and the relative high towing speeds of beam trawls. This study concludes that although a T90 codend reduces the size selectivity of flatfish, the lower SR of the T90 codend could be paired with a larger T90 codend mesh size for improved flatfish selectivity, and that a T90 mesh codend increases the size selectivity for commonly encountered roundfish in the Belgian beam trawl fishery.