A high-resolution, barotropic model of the North Atlantic is used to investigate bottom pressure variability on the Labrador and Newfoundland Shelves. The model has a free surface and is forced by surface atmospheric pressure and wind stress derived from twice-daily analyses from the European Centre for Medium-Range Weather Forecasts. Model output is compared with a yearlong set of bottom pressure data (1985--1986). Coherence squared between model and observations is significant at the 95% confidence level at almost all stations and exceeds 0.6 over most of the inner shelf, with good agreement in the phase. The model shows a tendency to overestimate the autospectral energy at periods less than 10 days but to underestimate the energy at longer periods. The model does less well at the outer shelf, where coherence squared is only marginally significant. The weaker performance of the model near the shelf break may indicate a role for physical processes not accounted for in the model (e.g., eddies, stratification) or, alternatively, might be due to the low signal-to-noise ratio in the bottom pressure measurements made at the shelf break. Model/data comparisons show that wind forcing dominantes over pressure forcing, except at the northern end of the Labrador Shelf, where both forcings are important. Model experiments run with and without Hudson Bay in the model domain demonstrate that Hudson Bay influences the Labrador Shelf in the 2 to 10-day period range. ¿ American Geophysical Union 1995