Abstract. It is desired to control excessive reactive nitrogen (Nr) deposition due to its detrimental impact on ecosystems. Using a three-dimensional atmospheric chemical transport model, GEOS-Chem, Nr deposition in the contiguous US and eight selected Class I areas (Voyageurs (VY), Smoky Mountain (SM), Shenandoah (SD), Big Bend (BB), Rocky Mountain (RM), Grand Teton (GT), Joshua Tree (JT), and Sequoia (SQ)) is investigated. First, modeled Nr deposition is compared with National Trends Network (NTN) and Clean Air Status and Trends Network (CASTNET) deposition values. The seasonality of measured species is generally well represented by the model (R 2 > 0.6), except in JT. While modeled Nr is generally within the range of seasonal observations, large overestimates are present in sites such as SM and SD in the spring and summer (up to 0.6 kg N ha month −1 ), likely owing to model high-biases in surface HNO 3 . The contribution of non-measured species (mostly dry deposition of NH 3 ) to total modeled Nr deposition ranges from 1 to 55 %. The spatial distribution of the origin of Nr deposited in each Class I area and the contributions of individual emission sectors are estimated using the GEOS-Chem adjoint model. We find the largest role of long-range transport for VY, where 50 % (90 %) of annual Nr deposition originates within 670 (1670) km of the park. In contrast, the Nr emission footprint is most localized for SQ, where 50 % (90 %) of the deposition originates from within 130 (370) km. Emissions from California contribute to the Nr deposition in remote areas in the western US (RM, GT). Mobile NO x and livestock NH 3 are found to be the major sources of Nr deposition in all sites except BB, where contributions of NO x from lightning and soils to natural levels of Nr deposition are significant (∼ 40 %). The efficiency in terms of Nr deposition per kg emissions of NH 3 -N, NO x -N, and SO 2 -S are also estimated. Unique seasonal features are found in JT (opposing efficiency distributions for winter and summer), RM (large fluctuations in the range of effective regions), and SD (upwind NH 3 emissions hindering Nr deposition). We also evaluate the contributions of emissions to the total area of Class I regions in critical load exceedance, and to the total magnitude of exceedance. We find that while it is effective to control emissions in the western US to reduce the area of regions in CL exceedance, it can be more effective to control emissions in the eastern US to reduce the magnitude of Nr deposition above the CL. Finally, uncertainty in the nitrogen deposition caused by uncertainty in the NH 3 emission inventory is explored by comparing results based on two different NH 3 inventories; noticeable differences in the emission inventories and thus sensitivities of up to a factor of four found in individual locations.