A new method of determining variations in the solar UV output is presented here. The method is based on the fact that the semi-diurnal (12 h) atmospheric tide is dominated by the component forced by ozone absorption of UV. This tide shows very little interannual variability; hence a variation in UV of 20%, as suggested by Heath and Thekaekara1, would have an easily discernable effect on the tide. It can be shown that a change of 20% in UV would result in a 12% change in the semi-diurnal tide. The failure to find any solar cycle-related variation in the semi-diurnal tide suggests that the UV varies by <2% over the sunspot cycle.