In the retrieval of satellite occultation experiments the optical extinction by atmospheric gases can be adequately modeled as the product of an extinction cross section with the number density. For aerosols the situation is more complex. Usually, the extinction spectrum is modeled by some analytical function of wavelength, controlled by a small number of parameters. However, a consensus about which function to use does not exist. The goal of this paper is to find out which analytical function is most suitable and how many parameters are necessary to capture the background aerosol extinction behavior. To do this, we first retrieve atmospheric constituent altitude profiles from a set of simulated spectra, measured by a virtual instrument of which the characteristics are based on the ones from the Global Ozone Monitoring by Occultation of Stars (GOMOS) instrument. Several aerosol extinction models are used in the retrieval. The effect of the aerosol model choice on the retrieval error and bias is then calculated by comparing the retrieved profiles with the ones that were used in the simulation. The results show that the use of a second-degree polynomial leads to good retrievals of aerosol extinction coefficients, while simpler models already give very acceptable results for the other species.