The selective properties of a bottom trawl fitted with a 70 mm diamond mesh codend and a 100 mm top square mesh panel (SMP) for hake (Merluccius merluccius), pouting (Trisopterus luscus and Trisopterus minutus) and red mullet (Mullus surmuletus) were investigated over the period 2011–2013. The experiments were carried out over three separate cruises aboard two commercial Basque bottom otter trawlers in the Bay of Biscay area. “Fall-through” experiments were also undertaken to estimate the potential size selection of 100 mm square mesh for the same species. Results from the “Fall-through” experiments and the at-sea selectivity cruises demonstrated that a 100 mm SMP has the potential to enable undersized and immature individuals to escape through the meshes. However, the selectivity cruises demonstrated that in practice, the SMP was largely ineffective at releasing undersized individuals as only a small fraction of the fish entering the trawl attempted to escape through the SMP during their drift towards the codend. The fraction attempting to escape was quantified by the “SMP contact probability” and was less than 4% for hake and red mullet and less than 15% for pouting. Furthermore, for each species, the release potential for the diamond mesh codend was found to be significantly lower than the length-at-maturity and the legal minimum conservation reference size. On average, the proportions of the total catch of undersized individuals of each species retained by the gear, were 52%, 17% and 45% for hake, pouting and red mullet respectively. Based on our findings, we conclude that the gear currently deployed by the Basque bottom otter trawl fleet operating in the Bay of Biscay is largely ineffective at releasing undersized hake, pouting and red mullet. The introduction of the obligation to land all catches, under the 2013 reform of the EU Common Fisheries policy will create new challenges for the Basque bottom otter trawl fleet and thereby an incentive to improve selectivity to avoid unwanted catches of undersized individuals.