Effective rangeland management requires careful consideration of the possible cumulative effects of different management options prior to making major management decisions. State-and-transition (S/T) models, based on ecological sites, capture our understanding ecosystem functioning and can be used to predict the potential cumulative impacts of different management regimes.  However, management units, such as grazing allotments or ranches, contain many ecological sites.  Thus, using S/T to assess cumulative effects at the scale of the allotment or ranch or larger has been difficult, if not impossible.  We have developed a dynamic modeling interface, the Cumulative Effects Analysis (CEA) Tool, which allows for S/T models to be used for cumulative impact assessment across these large landscapes. We will present this tool, and demonstrate its use across two large (>125,000 acres each) Bureau of Land Management (BLM) study areas in southern Idaho. We developed a set of S/T models based on groupings of similar ecological sites, and mapped the model states from field samples and satellite imagery. We defined a range of management scenarios related to juniper encroachment in one area and post-fire restoration in the other. Using repeated Monte Carlo simulations, we used the CEA Tool to estimate the range of potential future outcomes for each management scenario and the uncertainty in the projections. This ecological-site-based assessment allowed us to weigh the effectiveness of different scenarios in meeting management objectives.