The utility of the thermal wind equation (TWE) in relating tropospheric (850 to 300 hPa) wind and temperature on climatological time scales is assessed, based on data from 59 radiosonde stations in the western tropical Pacific during 1979-2004. Observed long-term mean and seasonal variations closely obey geostrophic balance; incorporating additional (ageostrophic) terms yields negligible improvement. We conclude that observed winds offer a useful constraint on the horizontal structure of monthly and longer temperature variations (although the reverse is not true close to the equator where f --> 0). This conclusion is also supported by general circulation model output. Wind data show a slowing of the midlatitude jets in the maritime continent region since 1979, indicating that tropical thicknesses and temperature have increased less than those poleward of 25°N/S. This pattern is consistent with Microwave Sounding Unit temperature trends in the region, but is exaggerated south of the equator in trends obtained directly from the temperature data. The latter are however sensitive to which stations are used and how the data are averaged, and the discrepancy is fairly small in a homogenized climatology from the Hadley Centre (HadAT). The discrepancy is most easily explained by spurious cooling at stations in the near equatorial western Pacific. These results support the use of the wind field as a way of overcoming heterogeneities in the temperature records in the monitoring of climate change patterns.