In this paper we evaluate two different methods for determining aerosol surface area and volume densities from multiwavelength extinction measurements and apply them to the Polar Aerosol and Ozone Measurement (POAM II) and Stratospheric Aerosol and Gas Experiment (SAGE II) data. Systematic and random components of the retrieval error are calculated as a function of the size distribution parameters for a unimodal lognormal size distribution using principal component analysis and an empirical formulation based on the 1 μm extinction measurement. The trade-off between these two components of the error is demonstrated for principal component analysis by varying the degree of constraint. We show how the systematic error can exceed the random error for lognormal size distributions typical of background aerosols. Retrieved surface areas for background aerosols can be underestimated by up to 50% through principal component analysis. Random errors depend on the estimated measurement error at each wavelength but, for the instruments considered, are expected to be 15--25%. Retrieved volume densities for background aerosols are underestimated by up to 30% through principal component analysis and suffer an additional random error component of 10--15%. The empirical formulation leads to altitude dependent systematic errors since the relative error depends on the particle number density. When different techniques are used to obtain aerosol surface area and volume, there is often a greater disagreement between different methods applied to the same data than the same method applied to data recorded by two different instruments. Systematic errors arising from the retrieval methodology should therefore be carefully considered in both validation and aerosol climatology studies. ¿ 1999 American Geophysical Union