In this study a theoretical approach to the analysis of branch plant location is developed and evaluated against data on Swedish multiplant firms and empirical evidence in the external control literature.The basic tenet of the analysis is that a spatial separation of ftrm activities implies an increase in the time-costs of coordination and control. The existence of these decentralization costs will, to a large extent, determine the location of activities in multi plant firms. Firstly, the establishment and survival of a branch plant will depend on the existence of a locational advantage vis-a-vis the headquarters. Secondly, there will be a differentiation between headquarter plants, characterized by activities with high decentralization costs, and branch plants, characterized by activities with low decentralization costs.The first proposition is supported by the distance-decay pattern found in branch plant location, the greater sensitivity of branch plants to place-bound location factors, the fmding that closure rates are higher for branch plants and by the positive effect of locational advantages on branch plant survival. Furthermore, the concentration of salaried employees and R&D functions in headquarter plants, the homogenization of branch plant activities and the high incidence of small branch plants located close to the headquarters may be interpreted as an adaptation to differences in decentralization costs.Finally, a theoretical model for the spatial distribution of industrial linkages is developed, and it is shown that branch plants, as parts of larger firms, may be expected to have less local linkages. This tendency is, however, difficult to demonstrate empirically because some branch plants are established to take advantage of local markets or input supplies.