Accurate spectral measurements of solar ultraviolet radiation are essential to an understanding of the photochemical and biological effects of ozone depletion. The analysis of spatial and temporal variations in ultraviolet fluxes will depend on the collation of spectra from many independent laboratories. However, results from diverse instruments operated in isolation may not be consistent with each other. To investigate the compatibility of different designs of spectroradiometer, a blind trial of six distinct instrument types was carried out at a suburban site in Greece. Comparisons were performed in daylight, and with tungsten lamps indoors. Excellent agreement was obtained in the relative spectral response of the instruments, but their absolute lamp calibrations varied, and did not generally agree with results from the daylight experiments. Simultaneous spectral scans by all instruments revealed discrepancies attributable to stray light, bandwidth, and cosine response, which would not otherwise have been apparent. The relative merits of the instrument characteristics are discussed.