UVB radiation plays an important role in tropospheric photochemistry since it determines the rate of ozone photolysis J(O1D) and subsequent formation of OH radicals. Consequently, changes of UVB radiation, for example due to changes of the stratospheric ozone amount, could alter the concentration of reactive tropospheric gases including ozone. An observation-based attempt is made to quantify the effect of changing UVB radiation on surface ozone peaks on a day-to-day scale using a time series of measurements at a Swiss mountain site. Seven years data of ozone, NO, NOx, and meteorological measurements from Chaumont (1140 m above sea level (asl)), total ozone and UVB measurements from Arosa (1847 m asl), and surface albedo from satellite observations are investigated. The study is restricted to fair weather days with moderately high NOx concentrations. Multiple regression analysis is performed using chemical, meteorological, and UV dependent variables to predict afternoon ozone peaks. From autumn to spring, positive deviations of ozone peaks are clearly connected with positive UVB deviations. The relation is statistically significant only in part of the seasonal data subsets; however, it is consistent with model studies. The estimated net effect on ozone peaks is normally within a range of 4 ppb, a range of about 6 ppb is predicted for large UVB changes. Applying the coefficients for the large interannual variability of the stratospheric ozone layer observed in spring in the last 10 years results in a range of variation of at most 1 to 1.5 ppb for monthly mean ozone peaks. For trends of J(O1D) from 1970 to 1990, a trend bias of surface ozone peaks on polluted fair weather days of less than 0.12 ppb/yr is calculated. Although the numbers are rather small, they may play a role in certain circumstances.