Two UV-Biometer 501A instruments were used to estimate global erythemal irradiance at two locations in southwest Sweden; the Earth Sciences Centre, University of Goteborg (57.69 deg. N; 11.92 deg. E) and the island of Nordkoster, 200 km to the north (58.83 deg. N; 10.72 deg. E). A semi-empirical radiative transfer model was used to calculate the global erythemally effective irradiance under clear skies. A ratio of the hourly measured to clear-sky modelled irradiance was then derived for zenith angles 35-70 deg. . Subsequent comparisons were then made with routine measurements of sunshine duration at Goteborg and sunshine duration, cloud cover, type and height at Nordkoster. Cloud transmission of UV-B irradiance decreases with increasing solar zenith angle, with cloud attenuation being 8% stronger at Nordkoster Island for zenith angles >>;60 deg. . Transmission also decreases withincreasing cloud cover such that overcast cloud conditions reduce transmissions by an average of 75%. In addition, cloud type affects theamount of ground incident irradiant flux. Fractus cloud afforded theleast UV-B transmission (0.16), while cirrus filaments afforded the most (0.95). The spatial and temporal distribution of clouds appears tobe non-random. Under conditions of 1 to 3 octas, sky cover, clouds appear to be concentrated in line with the sensor and Sun on more occasions than that expected given a random cloud distribution. The samecloud cover condition also resulted in many instances of ground incident irradiance above clear-sky values. The presence of cumuliform clouds appears to increase the likelihood of the latter phenomena. [ABSTRACT FROM AUTHOR]