The thermal emission from a cold, dense molecular cloud peaks in the far IR, and the spectrum is rich in molecular lines in the submillimetre and millimetre bands. Observation of these bands is hindered, however, by atmospheric water vapour, which absorbs the incoming radiation. Ground-based mm observations from Australia, where the atmospheric water vapour content typically contains ~10 mm precipitable (ppt) H2O, can only probe a few of the molecular transitions from the heavier molecules, such as CO, CS, HCO+ and HCN. Sub-mm observations would enable the higher rotational lines from many of these molecules to be studied, and open up other spectral features to scrutiny, such as the lines from hydrides (e.g. CaH, NH, SH) and neutral carbon at 370 and 610 μm. However, they cannot be made from Australia. While sites such as Mauna Kea, which has ~1 mm ppt H2O on the best days, open the sub-mm band to partial viewing, their utility is limited in comparison to the opportunities possible from the Antarctic Plateau. Here the column of H2O drops to 100–250 μm.