Photochemical reactions of trace compounds in snow have important implications for the composition of the atmospheric boundary layer in snow-covered regions and for the interpretation of concentration profiles in snow and ice regarding the composition of the past atmosphere. One of the prominent reaction is the photolysis of nitrate, which leads to the formation of OH radicals in the snow and to the release of reactive nitrogen compounds like nitrogen oxides (NO and NO2) and nitrous acid (HONO) to the atmosphere. We performed photolysis experiments using artificial snow containing variable initial concentrations of nitrate and nitrite to investigate the reaction mechanisms responsible for the formation of the reactive nitrogen compounds. Increasing the initial nitrite concentrations resulted in the formation of significant amounts of nitrate in the snow. A possible precursor of nitrate is NO2, which can be transformed into nitrate either by the attack of a hydroxy radical or the hydrolysis of the dimer (N2O4). A mechanism for the transformation of the nitrogen-containing compounds in snow was developed assuming that all reactions took place in a quasi-liquid layer (QLL) at the surface of the ice crystals. The unknown photolysis rates of nitrate and nitrite and the rates of NO and NO2 transfer from the snow to the gas phase, respectively, where adjusted to give an optimum fit of the calculated time series of nitrate, nitrite, and gas phase NOx with respect to the experimental data. Best agreement was obtained with a ~25 times faster photolysis rate of nitrite compared to nitrate. The formation of NO2 is probably the dominant channel for the nitrate photolysis. We used the reaction mechanism further to investigate the release of NOx and HONO under natural conditions. We found that the NOx emissions are by far dominated by the release of NO2. The release of HONO to the gas phase depends on the pH of the snow and the HONO transfer to the gas phase. However, due to the small amounts of nitrite produced under natural conditions, the formation of HONO in the QLL is probably negligible. We suggest that observed emissions of HONO from the surface snow are caused by the heterogeneous formation of HONO in the firn air. The reaction of NO2 on the surfaces of the ice crystals is the most likely HONO source.