Designing a reserve system for a threatened territorial species such as the Northern Spotted Owl requires the balancing of biological necessity against economic reality. The Spotted Owl conservation plan and the Pacific Northwest timber industry are in conflict because both demand large areas of mature and old‐growth coniferous forests. The primary focus of this paper is the efficient design of a reserve system for territorial species with obligate juvenile dispersal. We examine the relationship between the degree of aggregation of suitable owl habitat, the level of occupancy of that habitat by pairs of Spotted Owls, and the likelihood of owl persistence given different amounts and spatial arrangements of suitable habitat across the landscape. We develop a population model for Spotted Owls that includes an abstracted forest landscape where suitable owl habitat is arrayed in clusters embedded in a matrix not suitable for owl habitation. We conclude from our study of this model that, for any given fraction of the landscape set aside in reserves, the level of occupancy (efficiency of use) of that area will increase as the aggregation of suitable habitat increases. After the reserves reach a size that includes territories for 20 to 25 owl pairs, however, there are diminishing returns from further increases. Preserving connectivity and increasing the geographical extent of the reserve begin to outweigh increased size in importance in insuring the long‐term viability of the species.