Eleven vertical profiles of stratospheric NO3 have been obtained since 1992 using the AMON and SALOMON balloon-borne UV-visible spectrometers. The measurements are compared to the SLIMCAT 3D model and calculations based on the steady-state hypothesis for NO3. The calculations cannot reproduce some parts of the profiles which exhibit strong concentration fluctuations over few kilometres, as a consequence of the dependence of NO3 on local temperature variations. A statistical use of the data allows us to estimate the influence of the temperature dependence of the absorption cross-section on the data analysis, and the validity of the recommended reaction rates available in the literature. Discrepancies exist between the model based on recommended kinetics and observations at warmer temperatures. Nevertheless, the analysis is biased by local temperature inhomogeneities, and only a low-resolution vertical shape of the NO3 profiles can be retrieved.