A comparison of winds over the northern Gulf of Alaska found measured winds to be significantly weaker than calculated geostrophic winds. The latter required more than the standard 30% reduction to correct for frictional effects in the maritime boundary layer. An additional 11% is calculated based on principal axis analysis. A counterclockwise rotation of the instantaneous geostrophic wind vectors of 44¿ (59¿ total) beyond the standard 15¿ was required to align them with those measured at a coastal weather station. However, events in these two wind sets were well correlated throughout the year, and the rotary coherence was significant for frequencies lower than 0.895 cpd. On the basis of empirical orthogonal function analysis, most of the seasonal variability in this coastal current is associated with wind changes. Freshwater discharge is also important in driving the alongshore baroclinic flow and a secondary cross-shelf flow. ¿American Geophysical Union 1987