Abstract Radiometric measurements of terrestrial sunlight using three different types of broadband dosimeters were compared with equivalent integrated quantities obtained from simultaneous spectroradiometric measurements. Measurements were made at Durham, UK (55oN) during one day in mid-summer and one day in the autumn. By this means it was possible to encompass a wide range of ultraviolet irradiances. There was close agreement between UV-A irradiance measured using a broad-band radiometer and determined spectroradiometrically over the whole range of irradiances, when allowance was made for the spectral sensitivity of the UV-A radiometer. The agreement between erythemally-effective irradiance determined spectroradiometrically and the response of a Robertson-Berger meter showed some non-linearity due to the mismatch between the erythema action spectrum and spectral response of the sensor. There was a similar disparity in agreement between erythemally-effective dose determined spectroradiometrically and the response of polysulphone film for similar reasons. Nevertheless it is concluded that if these latter two dosimeters are calibrated using sunlight, or a solar simulator, as the source, they can yield data which are sufficiently reliable for many applications.