Abstract. In this work we consider large-scale set partitioning problems. Our main purpose is to show that real-world set partitioning problems originating from the container-trucking industry are easier to tackle with respect to general ones. We show such different behavior through computational experiments: in particular, we have applied both a heuristic algorithm and some exact solution approaches to real-world instances as well as to benchmark instances from the Beasley OR-library. Moreover, in order to gain an insight into the structure of the real-world instances, we have performed and evaluated various instance perturbations. 1