This paper presents a method for generating surface ultraviolet (UV) radiation maps over Europe, with a spatial resolution of 0.05¿, and potentially on a half-hour basis. The UV irradiance is obtained by interpolation in a look-up table (LUT), the entries of which are solar zenith angle, total column ozone amount, cloud liquid water thickness, near-surface horizontal visibility, surface elevation, and UV albedo. Both satellite (Meteosat, GOME) and nonsatellite (synoptic observations, meteorological model results, digital elevation model) data are exploited to assign values to the influencing factors. With the help of another LUT simulating the visible signal, Meteosat data are processed to retrieve the cloud liquid water thickness. The radiative transfer calculations are performed with the UVspec code. A preliminary step consists in generating an effective surface Meteosat albedo map from a series of 10 consecutive days. In this process the well-known difficulty of distinguishing clouds from snow-covered surfaces is encountered. An attempt is made to partially resolve the ambiguity by using the Meteosat infrared channel and modeled snow cover data. After additional empirical cloud filtering, the effective albedo map is used as a baseline to estimate the cloud liquid water thickness. The UV surface albedo is assigned uniform values for land and sea/ocean, except in the presence of snow. In this case it is given a value proportional to the Meteosat effective albedo. The total column ozone is extracted from the level 3 GOME products. The aerosol optical thickness is mapped by gridding the daily measurements performed by ~1000 ground stations. The digital elevation model is the GTOPO30 data set from the U.S. Geological Survey. European wide UV dose rate maps are presented for one day in April 1997, and the influence of the various factors is illustrated. A daily integrated dose map was also generated using 27 Meteosat acquisitions at half-hour intervals on the same day. The dose map produced in this way takes into account the evolution of the cloud field and is thought to be more accurate than if it were estimated from one data take, in particular at the relatively high spatial resolution of the product. Finally, a preliminary comparison of modeled dose rate and daily dose with measurements performed with a ground instrument is discussed. ¿ 2000 American Geophysical Union