Cirrus observations taken during EMERALD1 clearly showed vertical structure in ice crystal habit and size distribution. In this study, we use these observations along with radiative and cloud modelling to assess the importance of this observed vertical structure in size and habit on radiative properties and cloud evolution. We show that neglecting the vertical structure results in changes to the solar upwelling and downwelling fluxes of 10 and 8 W m -2, respectively. If further simplifications are made and aggregate and rosettes are treated more simplistically (as columns) then the upwelling and downwelling fluxes are altered by about 12 and 16 W m -2, respectively. [Correction added 8 May 2012 after original online publication: the end of the preceding sentence has been changed from '16 and 12 W m -2, respectively'.] It was noted that the effects of simplifying the vertical structure and habits act in competing ways in terms of the fluxes and therefore emphasize the need to consider both effects in order to improve modelling rather than considering either in isolation. Often, general circulation models have highly parametrized treatments of cloud properties. We illustrate that crude assumptions about ice crystal effective sizes used in models (from the literature) can result in fluxes being in error by 100 W m -2 or more compared to using the observations. The above results are contrasted with effects of ice crystal roughness (altering the asymmetry). We show that moderate estimates of roughness lead to changes in the upward and downward fluxes of about 10-15 W m -2. This shows that vertical structure, habits and knowledge of size distribution are key factors and on a par with such effects as ice particle roughness, which is considered to have a highly uncertain but important cirrus radiative effect. © 2011 Royal Meteorological Society.