One of the most important operational issues in an assembly line is to allocate workers to tasks (or workstations) so that their workloads can be well-balanced. We address a new workforce assignment problem in an assembly line where workers can perform all tasks but have a different skill level, which affects not only the throughput but also the number of defective products. We derive two mathematical programming models that can be used in various production environments. One is with the objective of the throughput maximization while keeping the defect rate under a given criteria and the other is for the minimization of the total rework time while achieving a given throughput. Several experiments are performed with the proposed two models by changing the number of workers and a proportion of skilled workers. We also compare the two models and provide managerial insights.