It has been known for decades that the mean circulation in the western Gulf of Mexico is a large anticyclone. This flow is driven both by wind stress curl and detached rings from the Loop Current. This paper is an attempt to examine their relative importance by focusing on the annual cycle of the observed flow. A new compilation of ship drift data shows that the flow along the western boundary is strongest in July and weakest in October. It appears physically plausible that the annual variation of this western boundary current is driven by the annual variation in wind curl augmented by Ekman pumping over the western gulf. Because the maximum in wind curl is so close to the western boundary, the phase delay between the annual cycles of curl and current is only approximately 1 month. The magnitudes of the curl and resulting current seem physically plausible. The response is baroclinic and penetrates to ~1000 m, the bottom of the main thermocline. The change in velocity at the sea surface is consistent with the cross-stream pressure gradient as determined by the changes in sea level at the coast and dynamic height changes offshore. The annual cycle is relatively large; at the annual minimum the flow is only approximately 25% of the maximum. The data suggest that this flow could be driven by wind curl, but the uncertainty is high. Consequently, the contribution of Loop Current rings turns out to be relatively small. The reason for this is that when rings reach the western wall, the steep bottom topography intersects the sloping fluid contours of the rings. The rings decay slowly by normal deepwater processes until they reach the edge of the continental shelf, and then they decay more rapidly by loss of fluid. Once the rings have reached the western boundary, the decay tine scale is found to be only ~70 days. ¿ American Geophysical Union 1993