The photolysis of nitrogen dioxide and formaldehyde are two of the most influential reactions in the formation of photochemical air pollution, and their rates are computed using actinic flux determined from a radiative transfer model. In this study, we compare predicted and measured nitrogen dioxide photolysis rate coefficients (jNO2). We used the Tropospheric Ultraviolet-Visible (TUV) radiation transfer model to predict jNO2 values corresponding to measurements performed in Riverside, California as part of the 1997 Southern California Ozone Study (SCOS’97). Spectrally resolved irradiance measured at the same site allowed us to determine atmospheric optical properties, such as aerosol optical depth and total ozone column, that are needed as inputs for the radiative transfer model. Matching measurements of aerosol optical depth, ozone column, and jNO2 were obtained for 14 days during SCOS’97. By using collocated measurements of the light extinction caused by aerosols and ozone over the full height of the atmosphere as model input, it was possible to predict sudden changes in jNO2 resulting from atmospheric variability. While the diurnal profile of the rate coefficient was readily reproduced, jNO2 model predicted values were found to be consistently higher than measured values. The bias between measured and predicted values was 17–36%, depending on the assumed single scattering albedo. By statistical analysis, we restricted the most likely values of the single scattering albedo to a range that produced bias on the order of 20–25%. It is likely that measurement error is responsible for a significant part of the bias. The aerosol single scattering albedo was found to be a major source of uncertainty in radiative transfer model predictions. Our best estimate indicates its average value at UV-wavelengths for the period of interest is between 0.77 and 0.85.