Observations from four, yearlong current meter deployments made in 1987/1988 near the shelf break in the Canadian Beaufort Sea are analyzed by statistical and spectral methods to characterize the structure and variability (vertical and horizontal) of the velocity field, and to compare measurements obtained during summer (ice-free) and winter (ice-covered) conditions. Three main frequency bands are defined: low frequencies (0.01--0.07 cpd), intermediate or synoptic-scale frequencies (0.07--0.7 cpd), and high frequencies (>0.7 cpd; including tides and inertial currents). Within this division, about 65% of the kinetic energy is in the low-frequency band, 15% in the intermediate band, and 20% is in the high-frequency band; the latter is dominated by tidal motions. Low-frequency variability is well correlated both vertically and horizontally along the shelf for separation distances exceeding 200 km. Variability in this band appears to be related to wind forcing and topographic wave propagation over large horizontal scales. Low-frequency variability is highest during October and November in association with the occurrence of intense autumn storms, and lowest during periods of ice cover due to wind shielding and increased frictional damping by ice cover. Intermediate-scale motions have significant correlation in the vertical, but poor correlation in the horizontal. Motions in this frequency band appear to be associated with mesoscale eddies advected by the mean flow. There is also evidence of local generation of intermediate-frequency current fluctuations by storms in late autumn and, again, an apparent damping during winter periods of ice cover. Finally, high-frequency motions show significant peaks in diurnal and semidiurnal tidal bands which are well correlated horizontally and vertically. However, other high-frequency motions in this band show no significant correlation in either the vertical or horizontal direction. There is no clear evidence in the current meter records of coupling between high-frequency motions and direct atmospheric forcing. ¿ 1998 American Geophysical Union