Just-in-time (JIT) production systems operate effectively if resources are correctly sized in terms of performance and availability. In JIT production lines in particular, performance can be considerably affected by machine reliability and the attendant maintenance policies. In this paper, a single-product, multi-stage manufacturing line with resource failures and controlled by a two card kanban system was modelled using discrete-event simulation. Several performance measures were evaluated to determine optimum operating policies given resource failures. In addition, different maintenance policies were implemented in order to better understand their impact on overall system performance. Several case examples were developed to further provide insight into the problem.