Using satellite data, the variability of a large number of stratospheric trace constituents can be estimated. These constituents need not themselves be measured by the satellite; their concentrations can be derived using photochemical steady-state relationships. The global coverage provided by the satellite over a long time period means that, for example, monthly zonal mean profiles can be derived. This has been done for H, OH, HO2, H202, C1, CIO, HC1, HOCI, C1ONO2, NO and O. The standard deviation of these quantities is a measure of their variability. We argue that comparing theoretical variability estimates with measurements is a better test of a photochemical theory than simply the comparison of single modelled and observed profiles.