Spatially explicit models that combine remote sensing with geographic information systems (GIS) offer great promise to land managers because they consider the arrangement of landscape elements in time and space. Their visual and geographic nature facilitate the comparison of alternative landscape designs. Among various activities associated with forest management, none cause greater concern than the impacts of timber harvesting on the composition, structure, and function of landscape ecosystems. A timber harvest allocation model (HARVEST) was used to simulate different intensities of timber harvest on 23,592-ha hypothetical landscapes with varying sizes of timber production areas and different initial stand age distributions. Our objectives were to: (1) determine the relative effects of the size of timber production areas, harvest intensity, method used to extract timber, and past timber harvest activity on the production of forest interior and edge; and (2) evaluate how past management (in the form of different initial stand age distributions) constrains future timber production options. Our simulations indicated that the total area of forest interior and the amount of forest edge were primarily influenced by the intensity of timber harvest and the size of openings created by harvest. The size of the largest block of interior forest was influenced most by the size of timber harvests, but the intensity of harvest was also significant, and the size of nontimber production areas was important when harvests were numerous and widely dispersed within timber management areas, as is often the case in managed forests. Stand age-class distributions produced by past harvest activity limited the amount of timber production primarily when group selection was used, but also limited clear-cutting when recent harvest levels were high.