In this work, the impact of aerosols over ultraviolet-B at the Earth's surface (UVB) was studied. The estimations of aerosols radiative effects on UVB are significant to UVB estimations, air quality studies, as well as assessments of the impact of regional environmental change. The study used simultaneously hourly values of UVB, global solar radiation (G) at the horizontal surface and direct normal solar radiation (In). In addition, other meteorological parameters such as air temperature (T), relative humidity (R) and cloudiness were collected. These data were measured at Qena, Egypt (26.20°N, 32.75°E, and 96 m amsl) through the period from 2001 to 2004. To analyze the relationship between UVB and aerosols, the dimensionless parameter UVB transmission (KtUVB) and Ångström turbidity coefficient (β) in cloudless conditions were considered. The results showed that there is no correlation between KtUVB and β in the wide range of SZA (3° > SZA < 75°). The change of KtUVB to change β was equal to −0.078 (the correlation coefficient, R = −0.22, i.e. 5% of the variability of KtUVB was explained by β). So, sensitivity analysis of aerosol effect in KtUVB to SZA was employed. The relationship between KtUVB and β was determined for a narrow ranges of SZA (the range is equal to 1°) and a linear regression was fitted for each range of SZA. The ΔKtUVB/Δβ, accordingly the correlation coefficient (R), increases with the increasing SZA, which means KtUVB becomes more sensitive to β as SZA increase. Datasets for each narrow range of SZA, which showed a correlation between KtUVB and β (−R > 0.50), were selected to quantify the relationship between both parameters. These selected datasets just show the effect of aerosols in KtUVB when their UVB penetrating influence is more than the influence of other atmospheric factors such as ozone, i.e. the effect of aerosols is to induce a notable reduction in KtUVB. For the selected datasets, ΔKtUVB/Δβ varied from −0.05 to −0.21 and its average value was equal to −0.12. The resulting regression analysis showed that the determination coefficients of linear fit vary from 0.25 to 0.77, i.e. 25% to 75% of the variance in KtUVB was explained by β.