The immediate use of exact methods for solving vehicle routing problems often seems not quite obvious from practical point of view, as there are too restrictive assumptions in these models. Looking at the routing problems in a chemical firm it is yet shown, how theoretically developed methods can be most helpful, as it concerns practical needs. In a case study the Sweepalgorithm becomes a valuable part of a multiple-stage heuristic procedure. Within this scope it is of special interest that there is no assumption of well-known transportation demand at the beginning of the planning period. The procedure delivers remarkable gains against the past results. © 1985 Springer-Verlag.