Seasonal and geographical variations in tropical intraseasonal wind variance are described using bandpass filtered 850 and 150 mb wind time series derived from rawinsonde observations. Three bandpass filters, with central response periods of 31, 47, and 99 days, are applied to the daily time series. The intermediate filter is designed to isolate variance associated with the “40–50 day oscillation.” The spatial coherence of the bandpass filtered wind fluctuations is examined using complex eigenvector analysis. Comparisons are made of u and v variance and large-scale structure of filtered wind anomalies for each season and frequency band, with emphasis on the u component. At stations across the western Pacific the 47-day filtered u150 variance is nearly constant with season. The largest seasonal variability in 47-day filtered zonal wind variance is at 150 mb at stations along and to the north of the equator between Africa and Southeast Asia, and in the central Pacific. Compared to the u150 variance... Abstract Seasonal and geographical variations in tropical intraseasonal wind variance are described using bandpass filtered 850 and 150 mb wind time series derived from rawinsonde observations. Three bandpass filters, with central response periods of 31, 47, and 99 days, are applied to the daily time series. The intermediate filter is designed to isolate variance associated with the “40–50 day oscillation.” The spatial coherence of the bandpass filtered wind fluctuations is examined using complex eigenvector analysis. Comparisons are made of u and v variance and large-scale structure of filtered wind anomalies for each season and frequency band, with emphasis on the u component. At stations across the western Pacific the 47-day filtered u150 variance is nearly constant with season. The largest seasonal variability in 47-day filtered zonal wind variance is at 150 mb at stations along and to the north of the equator between Africa and Southeast Asia, and in the central Pacific. Compared to the u150 variance...