High altitude clouds such as Cirrus have a substantial impact on the global radiation budget and their climatic impact depends on their micro-structure. Clouds composed of small crystals with effective radii less than 16 μm have a cooling effect, whilst clouds made up of larger crystals have a warming effect (Lynch et al., 2002; p 397). The success of cirrus microphysical modelling depends largely on the right choice of microphysical parameters and the diffusivity of water vapour (Dv) is one such crucial parameter. In this study we have explored the sensitivity of cirrus microphysics to the specification of Dv in large eddy model (LEM) simulations of cirrus case studies. We analysed important processes where vapour diffusion plays a role in the evolving microphysics. Cirrus clouds form at high altitudes where the molecular mean free path and hence Dv is significantly larger than the ground level value. To date some LEMs (e.g. the UK Met Office LEM) do not consider the height-dependence of the diffusivity of water vapour. Although this may not pose much of a problem for warm clouds formed in the lower boundary layer, for high altitude cirrus clouds, using ground level values could affect the microphysical development. In this study we have shown that crystal growth rates, the ice water mixing ratios, crystal number concentrations, auto-conversion rates of ice particles to form aggregates, the deposition rates of water vapour on aggregates, and the long-wave radiative cooling rates depend sensitively on the choice of the diffusivity of water vapour. The most widely used empirical formulation on the height-dependence of the diffusion coefficient is that described by Pruppacher and Klett (1997) and is valid over a temperature regime between -40°C and 40°C. Since many cirrus clouds form at temperatures colder than -40°C, it becomes imperative to use sophisticated formulations for an accurate prescription of Dv for cirrus studies. In this first LEM study we have used a Lennard-Jones (L-J) model to estimate Dv and applied it to two cirrus case studies. This formulation is accurate and valid over cirrus forming altitudes and is effective even when temperatures are colder than -40°C. First, we have shown that the L-J model can be easily adapted within LEMs to study cirrus clouds and thence we examined the resulting microphysics through simulations with and without the L-J update. We observed that the overall microphysical development was sensitive to the choice of the diffusion coefficient of water vapour. We believe that this study will cirrus modellers worldwide who are often constrained by the availability of microphysical observational data. Copyright © 2007 Royal Meteorological Society.