A hierarchy of processes operating at different spatial and temporal scales form landscape pattern, and changes to the patterns can have impacts on habitats and forest dwelling species. Managing landscapes under the auspices of sustainable forest ecosystems and emulation of natural disturbance requires knowledge of the scale at which landscapes are patterned. To better understand the role of natural disturbance in two distinct ecoregions of the Ontario boreal forest in Canada, we used thematic landcover maps derived from satellite imagery to evaluate differences in the relevant scales and nested hierarchies of forest pattern between two regions: northwest Ontario (relatively undisturbed forest dominated by Pinus banksiana and little fire suppression) and northeast Ontario (long history of forest harvest consisting mostly of Picea mariana and fire suppression). We define a relevant scale as the spatial level where non-random patterning of landscape structure occurs. Similar nested hierarchies occurred in forest and wildfire disturbance classes in northwest Ontario. In contrast, the relevant scales detected in the northeast occurred at fewer levels and in the disturbance class, which was the result of forest harvest, at only one level. The differences detected between the two regions indicate that current forest harvest practices that focus on landscape alteration at a single scale are creating new landscapes that are different from the natural landscape.