Abstract. Besides isoprene, monoterpenes are the non-methane volatile organic compounds (VOCs) with the highest global emission rates. Due to their high reactivity towards OH, monoterpenes can dominate the radical chemistry of the atmosphere in forested areas. In the present study the photochemical degradation mechanism of β-pinene was investigated in the Jülich atmosphere simulation chamber SAPHIR (Simulation of Atmospheric PHotochemistry In a large Reaction Chamber). One focus of this study is on the OH budget in the degradation process. Therefore, the SAPHIR chamber was equipped with instrumentation to measure radicals (OH, HO2, RO2), the total OH reactivity, important OH precursors (O3, HONO, HCHO), the parent VOC β-pinene, its main oxidation products, acetone and nopinone and photolysis frequencies. All experiments were carried out under low-NO conditions ( ≤ 300 ppt) and at atmospheric β-pinene concentrations ( ≤ 5 ppb) with and without addition of ozone. For the investigation of the OH budget, the OH production and destruction rates were calculated from measured quantities. Within the limits of accuracy of the instruments, the OH budget was balanced in all β-pinene oxidation experiments. However, even though the OH budget was closed, simulation results from the Master Chemical Mechanism (MCM) 3.2 showed that the OH production and destruction rates were underestimated by the model. The measured OH and HO2 concentrations were underestimated by up to a factor of 2, whereas the total OH reactivity was slightly overestimated because the model predicted a nopinone mixing ratio which was 3 times higher than measured. A new, theory-derived, first-generation product distribution by Vereecken and Peeters (2012) was able to reproduce the measured nopinone time series and the total OH reactivity. Nevertheless, the measured OH and HO2 concentrations remained underestimated by the numerical simulations. These observations together with the fact that the measured OH budget was closed suggest the existence of unaccounted sources of HO2. Although the mechanism of additional HO2 formation could not be resolved, our model studies suggest that an activated alkoxy radical intermediate proposed in the model of Vereecken and Peeters (2012) generates HO2 in a new pathway, whose importance has been underestimated so far. The proposed reaction path involves unimolecular rearrangement and decomposition reactions and photolysis of dicarbonyl products, yielding additional HO2 and CO. Further experiments and quantum chemical calculations have to be made to completely unravel the pathway of HO2 formation.