A continual trend of ozone depletion has drawn much attention to the biologically harmful UV-B radiation from the Sun reaching the Earth's surface. Satellite remote sensing provides a sole means of monitoring the global distribution of surface UV-B. At present, there are very few inversion algorithms with sufficient accuracy and robustness for operational application. A new algorithm is proposed here to infer surface UV-B irradiance and UV erythemal dose rate from satellite observations. The method is based on a simple model that treats UV radiative transfer in three distinct layers: an absorbing layer of ozone, a scattering layer of molecules, cloud and aerosol particles, and a layer of the Earth's surface. UV-B irradiance and dose rate at the surface are determined by the transmittance of the ozone layer, which can be derived from TOMS total ozone measurements, and the reflectance of the scattering layer, which can be determined from any UV or visible channel outside of ozone absorption bands. The inversion algorithm developed here is very simple (a couple of analytical expressions) and contains a few parameters that can be readily obtained from satellites (except for aerosol variables). The performance of the algorithm is validated against the results of comprehensive radiative transfer modeling using a DISORT-based model. Under a wide range of conditions (clear, cloudy, and turbid atmospheres) the retrieved surface UV-B irradiance and erythemal dose rate from the simple inversion algorithm are nearly as accurate as those calculated from the DISORT-based model but require much less computation and input data. ¿ 2000 American Geophysical Union