Mean monthly wind speed at 70 m above ground level is investigated for 11 sites in Minnesota for the period 1995-2003. Wind speeds at these sites show significant spatial and temporal coherence, with prolonged periods of above- and below-normal values that can persist for as long as 12 months. Monthly variation in wind speed primarily is determined by the north-south pressure gradient, which captures between 22% and 47% of the variability (depending on the site). Regression on wind speed residuals (pressure gradient effects removed) shows that an additional 6%-15% of the variation can be related to the Arctic Oscillation (AO) and Niño-3.4 sea surface temperature (SST) anomalies. Wind speeds showed little correspondence with variation in the Pacific-North American (PNA) circulation index. The effect of the strong El Niño of 1997/98 on the wind speed time series was investigated by recomputing the regression equations with this period excluded. The north-south pressure gradient remains the primary determinant of mean monthly 70-m wind speeds, but with 1997/98 removed the influence of the AO increases at nearly all stations while the importance of the Niño-3.4 SSTs generally decreases. Relationships with the PNA remain small. These results suggest that long-term patterns of low-frequency wind speed (and thus wind power) variability can be estimated using large-scale circulation features as represented by large-scale climatic datasets and by climate-change models.