Ultraviolet radiation (UV) (290-400 nm) plays an important role for atmospheric chemistry and photochemistry, human health, and it also plays a vital role for the formation and destruction of ozone in the atmosphere.An increase in UV can accelerate the chemical and photochemical reactions in the troposphere, lead to high 03 production, especially industrial cities in summer.Then, it is important to study UV and its affecting factors, as well as UV long-term trends, especially for the Beijing megacity.Analyzing observational data for solar radiation, meteorological parameters and total ozone concentration from January 1990 to December 1991 in Beijing enabled the development of an empirical method for estimation of UV irradiance (UVI) in clear skies.Predicted values from the model agreed well with observations.The mean relative bias for 24 months was 1.9%.UVI in clear sky conditions from 1979 to 1998 was calculated, and its long-term variation showed a declining trend of-3.89%, which was accompanied by a decrease in total 03 of-5.75%, a decrease in scattering factor (S/D, the ratio of solar scattered to direct radiation) of-20.79%, and an increase in water vapor content of +4.76%.A minor but interesting phenomenon was found for UVI variation in summer.It is hypothesized that many substances in the atmosphere, including volatile organic compounds (VOCs) and their oxidation products (including secondary organic compounds, SOA), very fine particles and others absorb and/or utilize UV energy.The long-term UVI trends and its main controlling factors in four seasons during the previous 2 decades are discussed, UV energy consumption by atmospheric chemical and photochemical processes is especially important during summer.The minor but important phenomenon of UVI decrease in summer enlarges the effect of stronger UV absorption and utilization by gases, liquids, particles (GLPs) through OH radicals.Therefore, the measurements in the laboratory and field are needed, including UV absorption and utilization parameters of GLPs, then, new data and mechanisms should be introduced into the models, so as to improve our understanding and forecasting of UV radiation.