We consider a flexible job shop scheduling problem coming from printing and boarding industry. A memetic algorithm is proposed to find a set of on-dominated solutions for the makespan and the maximum lateness criteria. A memetic algorithm is a genetic algorithm hybridized with a local search. Our genetic algorithm is based on NSGA-II where the mutation has been replaced by a Tabu Search algorithm based on a linear combination of criteria. Computational results show the interest of the combination of the two approaches.