This work studies rearrangement problems involving the sorting of robots or objects in stack-like containers, which can be accessed only from one side. Two scenarios are considered: one where every robot or object needs to reach a particular stack, and a setting in which each robot has a distinct position within a stack. In both cases, the goal is to minimize the number of stack removals that need to be performed. Stack rearrangement is shown to be intimately connected to pebble motion problems, a useful abstraction in multi-robot path planning. Through this connection, feasibility of stack rearrangement can be readily addressed. The paper continues to establish lower and upper bounds on optimality, which differ only by a logarithmic factor, in terms of stack removals. An algorithmic solution is then developed that produces suboptimal paths much quicker than a pebble motion solver. Furthermore, informed search-based methods are proposed for finding high-quality solutions. The efficiency and desirable scalability of the methods is demonstrated in simulation.