<p>In this study, due date assignment problem is addressed in a dynamic and stochastic single machine environment with family setups. A due date is to be assigned for each new job immediately at the time of its arrival. While assigning a due date for the new arrival, non-complete jobs that have arrived before that job and whose due dates are already assigned, are also considered and rescheduled. While assigning a short due date for the new arrival as close as possible, compliance to assigned due dates of non-complete jobs is necessary. These two objectives conflict with each other. The trade-off between assigning tight and loose due dates reflects the trade-off between competitive power and delivery performance. To solve this problem, a two-phase solution methodology is proposed where a sequential lexicographic approach is followed. In the first phase, a capacity allocation takes place for families before observing any actual jobs arrivals, based on expected workload and arrival estimation, in a periodic and static manner. In this phase, families can be assigned to batches, and a batching structure is formed. In the second phase, a due date is assigned for the new arrival immediately in an online fashion, based on the outputs of the first phase. A mixed integer programming model is developed for each phase. Also, a heuristic algorithm is proposed for the second phase. Simultaneously, a discrete event simulation is carried out to imitate a real production system. The performance of the designed policy is measured through the developed simulation model and results are reported under different control and system parameters.</p>