Erythemal daily UV doses have been calculated using input data from the ERA-40 reanalysis for years 1958-2002. The quality of input parameters for UV calculations has been validated with available ground based total ozone data and estimates of Cloud Modification Factor (CMF), and the results have been compared with existing UV data. Owing to limited availability of validation data especially during 1960s and 1970s the analysis focused on sites located in central and northern Europe. The current work is the first step in deriving reliable long-term UV time series from the ERA-40 reanalysis. Total ozone from the ERA-40 reanalysis is affected by significant biases, especially before satellite ozone measurements were available for assimilation. Estimations of the effect of clouds on surface UV were made using global radiation (300-3000 nm) budgets at the surface because available ERA-40 cloud data do not allow good estimates of surface daily UV doses. There are some problems with ERA-40 solar radiation budgets which cause systematic biases in calculated daily UV doses. Comparison of calculated daily erythemal UV doses against ground-based UV data indicate that ERA-40 UV doses are typically overestimated by 6-18% in central and northern Europe and underestimated by 9-17% at Davos, Switzerland. Root-mean-square errors of the calculated daily UV doses are usually in the range of 30-40%. Trends of UV doses were calculated for the concurrent period of the TOMS satellite UV data (1979-2002). The trends of zonally averaged ERA-40 and TOMS UV agree well and are mostly of the same sign and magnitude.