Landscapes are often heterogeneous in abiotic factors such as topography, climate, and soil, yet little 15 known about how these factors may influence the spatial distribution of primary productivity. We report estimates of aboveground net primary productivity (ANPP) in 90 sample stands stratified by cover type and elevation class, and use the results to predict ANPP across a portion of the Greater Yellowstone Ecosystem. Tree ANPP was estimated -by sampling tree density by species and diameter classes and estimating average annual diameter increment by tree coring. Biornass for current tree diameter and past tree diameter were calculated by species and diameter class for each stand using the dimension analysis software BIOPAK. Shrub ANPP was estimated by calculating current biomass from basal area using BIOPAK and dividing by the assumed average life span of the shrubs. Clipping at the end of the growing season was used to estimate herb ANPP. Differences in ANPP among cover types and elevation classes were examined with analysis of variance. Multiple regression was used to examine relationships between ANPP, and soil parent material, topography, and cover type. The best regression model was used to predict ANPP across the study area. We found ANPP was highest in cottonwood, Douglas-fir, and aspen stands, intermediate in various seral stages of lodgepole pine, and lowest in grassland and sagebrush cover types. Parent material explained significant variation in ANPP in mature and old-growth lodgepole pine stands, with rhyolite ash/loess being the most productive parent material type. ANPP decreased with increasing elevation in most cover types, possibly because low temperatures limit plant growth at higher elevations in the study area. ANPP was not related to elevation in mature and old-growth lodgepole pine stands, due to relatively rapid growth of subalpine fir at higher elevations. A regression model based on cover type and elevation explained 59% of the variation in ANPP among the sample stands. This model was used to generate a spatially continuous surface of predicted ANPP across the study area. The frequency distribution of predicted ANPP was skewed towards lower levels of ANPP, and only 6.3% of the study area had a predicted ANPP level exceeding 4500 kg/ha/yr. Patches high in predicted ANPP were primarily at lower elevations, outside of Yellowstone National Park, and near the national forest/private lands boundary. These patterns of ANPP may influence fire behavior, vertebrate population dynamics, and other ecological processes. The results reinforce the need for coordinated management across ownership boundaries in the Greater Yellowstone Ecosystem. © 1998 Kluwer Academic Publishers. Abstract reproduced by permission.