Winds measured over the continental shelf off Vancouver Island during the summers of 1979 and 1980 are compared with winds computed by the National Maine Fisheries Service (Monterey, California), using synoptic surface atmospheric pressures. Although there is considerable qualitative similarity between the two sets of data, the computed 6-hourly winds fail accurately to resolve short-period fluctuations associated with transient wind systems and tend to underestimate the percentages of time of low and high wind speeds. In the frequency domain, the computed winds are representative of the oceanic winds for periods exceeding roughly 2 days but only marginally representative for periods of 1 1/2 -2 days. At periods less than about 1 1/2 days, computed winds are not reliable indicators of oceanic wind variability. Moreover, the computed winds possess a large-amplitude sea breeze (diurnal) component that is present in the inner shelf wind data but not in the outer shelf wind data. This sea breeze component in th computed winds originates through use of land-based atmospheric presure values and is not indicative of the offshore wind field. Spectra of the observed winds suggest a k-5/3 law behavior (k is the horizontal wave number) consistent with the existence of a mesoscale, two-dimensional, reverse cascading energy inertial range. Spectra for computed winds, on the other hand, are more consistent with a k-3 law appropriate to geostrophic turbulence and the existence of a macroscale, two-dimensional, cascading entrophy inertial range.