Atmospheric aerosols are among the most variable components of the Earth's atmosphere, which is one of the reasons they have largely been omitted from general circulation models. However, recent work has suggested that the shortwave radiative effects of aerosols can be quite significant, possibly rivaling current greenhouse gas forcing. There is, of course, no time to perform a realistic radiative transfer calculation at each grid point and time step. Instead, one must find a way to parameterize aerosol effects and how they may vary as the aerosol type and loading vary along with the humidity. Radiative perturbation theory is an ideal tool for the development of such parameterizations, as we have recently shown. In this paper, we briefly outline the principles of perturbation theory (the details can be found in the references) and show how it may be applied to this task. Then we examine the specific problem of computing heating rates, which has tested perturbation theory up until now. We show that far more accurate results may be obtained by taking the difference in fluxes either side of the atmospheric layer in question, each calculated using an exponential formulation of perturbation theory. ¿ American Geophysical Union 1996