Atmospherically adjusted coastal sea level observations (ASL) and wind measurements at stations along the east coast of North America from Cape Hatteras, North Carolina, to Eddy Point, Nova Scotia, for a 6-month winter period were used to study the spatial structure of ASL and wind stress on the continental shelf and to determine the low-frequency response of ASL to wind forcing ASL and wind stress were coherent over distances of 1300 km for motions with periods between 60 and 600 hours. Northward movement of storm systems and of the associated ASL response was observed for stations north of Cape May, New Jersey. South of Cape May the phase of ASL suggested southward movement of pressure disturbances. Londshore wind accounted for 55% of ASL energy and led ASL by 8--12 hours. Cross-shore winds were generally not coherent with ASL. In the MIddle Atlantic Bight (MAB), sea level response was 17 cm for a 1 dyn cm-2 longshore wind and symmetric in the longshore wind stress directions. In contrast, from Nantucket north, sea level response to longshore wind stress was weaker (9 cm for a 1 dyn cm-2 wind to the northeast). The response for wind to the northeast was twice as large as the response for wind stress toward the southwest. If coastal winds were representative of winds over the shelf, the observations imply that the wind-driven longshelf geostrophic currents were larger for unit wind stress in the MAB than in the Georges Bank-Scotian Shelf region. The regional change in ASL response may be due to the large semidiurnal tidal currents which occur in the Georges Bank region or to the rough topography on the Scotian Shelf, which effectively increases bottom friction. The regional change in ASL reponse suggests a divergence in longshore flow and an onshore flow on the shelf south of Nantucket for winds to the southwest. A low-frequency longshelf ASL slope was observed in the MAB that was incoherent with longshelf wind stress.