Multimesh nylon gillnets were set in three Ebro Delta (North-East of Spain) lagoons to determine mesh selectivity for the inhabiting fish community. Each gillnet consisted on a series of twelve panels of different mesh size (ranging from 5.0 to 55.0 mm bar length) randomly distributed. The SELECT method (Share Each Length's Catch Total) was used to estimate retention curves through five models: normal location, normal scale, gamma, lognormal and inverse Gaussian. Each model was fitted twice, under the assumptions of equal and proportional to mesh size fishing effort, but no differences were found between approaches. A possible situation of overfishing in the lagoons, where artisanal fisheries are carried out with a low surveillance effort, was assessed using a vulnerable species inhabiting these brackish waters as case study: the sand smelt, Atherina boyeri. The minimum size for its fishery has not been established, thus remaining under an uncontrolled exploitation situation. Therefore, a Minimum Landing Size (MLS) is proposed based on sexual maturity data. The importance of establishing an adequate MLS and regulate mesh sizes in order to respect natural maturation length is discussed, as well as, the proposal of other measures to improve A. boyeri fishery management.