This article addresses the twofold question of at which location and at what point in time waste is recycled. Immediate recycling is unnecessary since storage is available to a limited extent. One approach is to conduct large-scale recycling centrally at one location, but this would lead to high transport efforts for the waste. In this paper, a different, distributed approach is presented. Due to the miniaturization of production and recycling units and their being embedding in standard containers, it is possible to relocate them from site to site at short notice. We present a dynamic location problem for mobile recycling units (DLPMRU) to limit both the transport of waste and the movement of mobile recycling units. In addition to a mixed-integer programming (MIP) formulation, a bilevel decomposition algorithm is proposed, which, in an iterative process, determines the movement of recycling plants through several local search operations and a linear programming (LP) subproblem. Early computational tests show that our algorithm achieves competitive results, especially for large-sized problems.