Estimation of the degradation rate of plant protection products in water under realistic conditions may be important for correct estimation of exposure concentrations for regulatory purposes. Standardized tests for degradation in water and in water-sediment systems in the laboratory exist, but these do not reflect degradation under field conditions. This is especially true for studies on photolytic degradation; therefore, generally photolytic degradation is not accounted for in the lower tiers of the exposure assessment. The aim of this study is to develop a procedure for the estimation of photochemical degradation rates from outdoor cosm experiments for use in the higher tiers of the exposure assessment. Observations in outdoor ponds or cosms are regularly used as a higher-tier risk assessment to evaluate the ecotoxicological effects on the aquatic ecosystem in a more realistic way. By means of inverse modelling of the behaviour of the compound in the cosm we determined the degradation rate in water, DegT50, for three compounds which are known to degrade photolytically. We did so by coupling the fate model TOXSWA to the optimisation tool PEST and determined the DegT50 for a daily reference UV radiation, weighed with a vitamin-D action spectrum, assuming that the degradation rate was directly proportional to the amount of weighed UV radiation. The UV radiation data were derived from satellite-based observations accounting for the effects of the thickness of the ozone layer and the cloud cover. For cosm studies with metribuzin, imidacloprid and metamitron we obtained satisfactory estimates of DegT50 values. After correcting these for effects of water depth, coverage of water surface by plants and the skyview factor on the UV radiation in the water, the variation between the DegT50 values of metribuzin and imidacloprid was smaller than the variation between DegT50 values derived (in an earlier study) by assuming that the degradation rate depended on water temperature and not on UV radiation. This indicates that ...