We present initial results from a water-vapour monitoring campaign initiated in 1996. A certain amount of caution must still be applied to the interpretation of the water-vapour statistics as a large majority of the data have been taken in summer, or autumn. Continued monitoring is accumulating strong evidence of a seasonal effect, with typically much smaller water-vapour columns during the winter months. The data suggests that a high percentage of nights (/~39%) are acceptable for thermal infrared observing (/<3 mm of water vapour at the zenith), even during the poorest epochs of the year, while this figure increases to /~70% during the winter months. A significant percentage of the time (/~10%), independently of season, the water vapour column is below 1 mm at the zenith; such small water-vapour columns may remain stable for several days at a time.