We study the statistical properties of the coastal wind field along the west coast of North America for two summers, 1981 and 1982 (Coastal Ocean Dynamics Experiments 1 and 2, respectively), and the intervening winter using measured winds and geostrophic winds calculated from Fleet Numerical Oceanography Center atmospheric pressure analyses. Summer wind fluctuations are driven primarily by the interaction between two relatively stationary pressure systems, the North Pacific subtropical high and southwest U.S. thermal low, and by their interactions with propagating atmospheric systems to the north. In particular, propagating cyclones and associated fronts are often followed by a northeastward intensification of the high, producing strong upwelling events along the California coast. This summer event sequence occurs more frequently and is displaced farther to the south on average during summer 1981. Winter wind fluctuations are primarily driven by propagating cyclones and anticyclones, and they tend to have larger variance and space scales than in summer. A preference for poleward (equatorward) propagation exists in summer (winter), and the largest time scales are observed in summer 1982. Coastal atmospheric boundary layer processes substantially modify winds within 100--200 km of the coast. Consequently, measured wind fluctuations are strongly polarized in the alongshore direction and have means and rms amplitudes that can vary considerably between nearby stations along the coast. Calculated wind fluctuations are less polarized in the alongshore direction and have alongshore correlation scales about 60% larger than those for measured winds. They represent fluctuations with alongshore wavelengths of ≥900 km rather well but represent poorly those with smaller wavelengths and those due to coastal atmospheric boundary layer effects. ¿American Geophysical Union 1987