Marine birds have the potential to be affected by human activities in the ocean environment such as offshore wind energy development. This report describes a project that developed maps of the spatial distributions of marine bird species in U.S. Atlantic Outer Continental Shelf (OCS) waters that can be used to aid ocean planning in the region and guide future data collection efforts. Sighting survey data from over three decades contained in the ‘Northwest Atlantic Seabird Catalog’ database, along with Eastern Canada Seabirds at Sea data from Canadian Wildlife Service, Environment and Climate Change Canada, were analyzed to derive seasonal maps of the spatial distributions of 47 marine bird species in U.S. Atlantic OCS and adjacent waters from Florida to Maine. Spatial predictive modeling was applied to the survey data to account for spatial and temporal heterogeneity in survey effort, platform, and protocol. An ensemble machine-learning technique, component-wise boosting of hierarchical zero-inflated count models, was used to relate the relative density of each species to multiple spatial and temporal predictor variables while accounting for survey heterogeneity and the aggregated nature of sightings. Dynamic spatial environmental predictor variables were formulated as long-term climatologies. The modeling technique allowed for complex non-linear relationships between response and predictor variables and interacting effects among predictors. Bootstrapping was used to derive estimates of the uncertainty in model predictions. Model predictions are presented as seasonal maps of the relative density of each study species throughout the study area. The maps were reviewed by experts with experience and knowledge of marine birds in the study area and their comments were incorporated in this report. The maps indicate where species are likely to be more or less abundant. The analysis was not designed to estimate the actual number of individuals/density of a given species that would be expected in any location, so the maps should not be interpreted that way. Also, the maps represent the spatial distributions of birds averaged over time (e.g., across days within a season and across years for a given season). The analysis was not designed to provide predictions of the density of birds that would be expected in a specific location at a specific date or time, so the maps should also not be interpreted that way. Two indications of the uncertainty associated with the model predictions are provided. First, a hatched overlay is included on the maps of predicted relative density to indicate areas with no survey effort. Model predictions in areas with no survey effort should be interpreted with extreme caution. Predictions in these areas were often questionable or unrealistic, so we recommend additional field surveys in these areas to validate the model predictions. Second, estimates of the precision of model predictions are presented as maps of the coefficient of variation (CV) of predicted relative density. Less precise predictions (i.e., higher CV) should be interpreted with more caution. The maps of predicted relative density should always be considered in conjunction with these two indications of uncertainty. The relative importance of different predictor variables is also presented, indicating which variables most influenced the predicted distributions for each species in each season. While the primary objective of this study was not to determine the ecological drivers and mechanisms behind the spatial distributions of marine bird species in the study area, our model results may provide useful hypotheses for future studies aimed more at ecological inference.