We present an analysis of observations of the auroral/polar regions of Saturn, carried out in 1999, 2004 and 2005, making use of the facility spectrometer, CGS4, on the United Kingdom Infrared Telescope (UKIRT), Mauna Kea, Hawaii. We obtain temperatures of 380 ( ± 70 ) K in 1999 and 420 ( ± 70 ) K in 2004. (The 2005 data has insufficient spectral resolution for a temperature determination to be made.) Our most probable interpretation is that the temperature of Saturn's auroral/polar H + 3 layer should be taken as 400 ( ± 50 ) K. This is lower than the value obtained by Miller et al. [Miller, S., and 10 colleagues, 2000. Philos. Trans. R. Soc. 358, 2485–2502], which is shown to be in error. Our analysis reveals clearly that the line emission due to H + 3 varies considerably, showing nearly an order of magnitude increase when one compares the data obtained in 1999 with those of 2004. Our conclusion is that this variability is (mainly) due to the changing H + 3 column density. By analogy with modelling results obtained for Jupiter, we estimate that the particle (keV electron) precipitation experienced by Saturn must be ∼20 times greater in 2004 than in 1999, to produce this additional ionisation. The H + 3 emission increases, but this is insufficient to offset most of the heating due to the extra particle precipitation, indicating that this ion does not act as a “thermostat” on Saturn, in the same way that it does on Jupiter.