Sea level data from St. John's, Newfoundland, and from Nain, Labrador, are compared with results from a 1/4¿¿1/4¿ resolution numerical model of the Newfoundland and Labrador shelf and the neighboring Labrador Sea. The model is barotropic, employs dynamics linearized about a state of rest, and uses linear bottom friction. The model is driven by the seasonal part of the Hellerman and Rosenstein wind stress field and by inflow from the rest of the North Atlantic specified along the eastern boundary. The latter is taken from a 1¿¿1¿ version of the model applied to the North Atlantic between 10 ¿S and 80 ¿N. The results indicate that our simple model can account for that part of the annual cycle of sea level on the shelf directly attributable to wind forcing. In particular, the model can account for the difference between observed monthly mean sea level at St. John's, corrected for atmospheric pressure variations, and the seasonal steric height anomalies above 150 dbar on the neighboring shelf. We also compare our model results with the linear regression analysis carried out by Thompson et al. on monthly mean sea level data from Nain for the period 1964 to 1972. The model confirms the gains calculated for both the local wind driven signal and also that part of the signal due to wind forcing over the North Atlantic beyond the Labrador Sea. Comparing St. John's with Nain, we find that at St. John's, the model response is accounted for primarily by wind set-up on the shelf, whereas at Nain, the influence of North Atlantic wind forcing is a larger part of the total signal. These results suggest that data from the tide gauge at Nain could be useful for verifying model-predicted changes in the large-scale ocean circulation. The results also seem to be confirming the presence of the offshore transport variations predicted by the model. ¿ American Geophysical Union 1990