The assessment of the climatic effects of an aerosol with a large variability like mineral dust requires some approximations whose validity is investigated in this paper. Calculations of direct radiative forcing by mineral dust (short-wave, long-wave and net) are performed with a single-column radiation model for two standard cases in clear sky condition: a desert case and an oceanic case. Surface forcing result from a large diminution of the short-wave fluxes and of the increase in down-welling long-wave fluxes. Top of the atmosphere (TOA) forcing is negative when short-wave backscattering dominates, for instance above the ocean, and positive when short-wave or long-wave absorption dominates, which occurs above deserts. We study here the sensitivity of these mineral forcings to different treatments of the aerosol complex refractive index and size distribution. We also describe the importance of the dust vertical profile, ground temperature, emissivity and albedo. Among these parameters, the aerosol complex refractive index has been identified as a critical parameter given the paucity and the incertitude associated with it. Furthermore, the imaginary part of the refractive index is inadequate if spectrally averaged. Its natural variability (linked to mineralogical characteristics) lead to variations of up to ± 40% in aerosol forcing calculations. A proper representation of the size distribution when modelling mineral aerosols is required since dust optical properties are very sensitive to the presence of small particles. In addition we demonstrate that LW forcing imply a non-negligible sensitivity to the vertical profiles of temperature and dust, the latter being an important constraint for dust effect calculations.