A barotropic, shallow-water model of the North Atlantic is used to investigate variability in adjusted sea level on time scales of a few days to a few months (by “adjusted,” we mean that the inverse barometer is removed from both the model-computed sea level and the observations). The model has 1/3° × 0.4° resolution in latitude and longitude, respectively, and is forced using atmospheric pressure and wind stress data derived from European Centre for Medium Range Weather Forecasts (ECMWF, 1994) analyses. The model results are compared with coastal tide gauge data. Along the western boundary, from St. John's, Newfoundland, to Fernandina Beach, Florida, coherence squared between model and data is greater than 0.5 in the period range 3 to 10 days. South of Cape Hatteras, the model underestimates the amplitude seen in the data, with much better agreement north of the Cape. Model performance on the eastern boundary is generally poor. We suggest this is because on the eastern boundary, the shelf width is much narrower, compared to the internal radius of deformation, than on the western boundary. In addition, the model resolution is insufficient to adequately represent the shelf on the eastern boundary. The poorer agreement south of Cape Hatteras may be due Gulf Stream effects not accounted for by the model dynamics. Finally, we discuss the model-computed variability in the ocean interior.