The major factors causing differences between satellite-derived and ground-based ultraviolet (UV) erythemal irradiances and doses are discussed. Measurements totaling more than 4700 days during 1997--1999 were obtained at 8 stations (22¿S--64¿S) of the Argentine UV Monitoring Network. The satellite retrieval uses radiative transfer calculations for cloud- and aerosol-free conditions multiplied by correction factors for clouds and aerosols. Key parameters are total ozone, cloud optical depth, and surface albedo derived from Total Ozone Mapping Spectrometer (TOMS). When no aerosol correction is applied, systematic differences of satellite-derived erythemal irradiance relative to ground-based measurements amount to +1% at a tropical high-altitude Andean location, +10% at stations in the central Pampas, +5% at southern Patagonian sites, and -7% at the southernmost continental and Antarctic stations with varying snow cover. When an aerosol correction is applied by estimating minimum and maximum aerosol loading, the systematic differences are within ¿10% for all snow-free stations. To reduce the differences at places with varying snow conditions, an average surface-albedo climatology must be used instead the TOMS climatology of minimum albedo. Although the statistical uncertainty of the differences increases with TOMS reflectivity, the systematic difference is independent of TOMS reflectivity for most of the stations, so on average the comparison for cloudy situations is as good as for clear-sky conditions. The comparison for daily erythemal doses gives similar results with smaller statistical uncertainty. Measured uncertainties are in agreement with a theoretical analysis. For most locations, well-characterized ground-based instruments should agree with TOMS satellite estimations within 10% if aerosol corrections are known.