UV radiation at 4 wavelengths (305, 320, 340 and 380 nm) and PAR were measured from May 1994 to October 1999 using Biospherical UV-radiometers. A surface reference sensor located on the roof of the Marine Station at Helgoland recorded values every 5 min, and an equivalent profiling underwater sensor was used for measurements in the sea at approximately monthly intervals. The ratio of 305 nm radiation to PAR varied seasonally, with a 14-fold increase from winter to summer. A much weaker seasonal trend (c. 1.5-fold) was apparent in the ratio of 320 nm radiation to PAR, but there was no seasonal trend in the ratios of 340 or 380 nm radiation to PAR. The year-to-year variations in 305 nm radiation were also much greater relative to PAR than for the other UV wavelengths, but there was no evidence of a change in the 305:PAR ratio over the study period. The ratios of both 305 and 320 nm radiation to PAR increased from dawn to midday, but those of 340 and 380 nm radiation were almost constant through the day, except shortly before sunrise and after sunset when the proportions of 340 and 380 nm radiation increased. Underwater measurements of PAR and UV suggest that the 1% depth for 305 nm radiation was little more than 1 m, but this estimate is valid only for summer and autumn because, in other seasons, few reliable readings for 305 nm radiation could be obtained underwater, and no attenuation coefficient could be calculated. The 1% depths recorded for the other UV wavelengths in the middle 6 months of the year were 2.0 m for 320 nm, 2.6 m for 340 nm and 4.6 m for 380 nm, compared with 12 m for PAR, but the attenuation of all wavebands increased sharply in October and remained higher until March. An analysis of the influence of sun angle, total column ozone concentration, the proportion of skylight, and cloud cover on the ratio of UV-wavelengths to PAR in surface irradiance demonstrated that solar angle has a greater influence than ozone concentration on the irradiance at 305 nm, and that the typical occurrence of ozone "holes" in spring may not result in higher UV-B irradiances than occur under higher ozone concentrations in summer. The implications of the data for attempts to model the biological effects of natural UV-radiation on marine organisms are considered.