The reaction CH3C(O)O2 +HO2 → CH3C(O)OOH+O2 (Reaction R5a), CH3C(O)OH + O3 (Reaction R5b), CH3 + CO2 + OH + O2 (Reaction R5c) was studied in a series of experiments conducted at 1000 mbar and (293±2) K in the HIRAC simulation chamber. For the first time, products, (CH3C(O)OOH, CH3C(O)OH, O3 and OH) from all three branching pathways of the reaction have been detected directly and simultaneously. Measurements of radical precursors (CH3OH, CH3CHO), HO2 and some secondary products HCHO and HCOOH further constrained the system. Fitting a comprehensive model to the experimental data, obtained over a range of conditions, determined the branching ratios α(R5a) = 0.37±0.10, α(R5b) = 0.12±0.04 and α(R5c) = 0.51±0.12 (errors at 2σ level). Improved measurement/model agreement was achieved using k(R5) = (2.4±0.4)×10−11 cm3 molecule−1 s−1, which is within the large uncertainty of the current IUPAC and JPL recommended rate coefficients for the title reaction. The rate coefficient and branching ratios are in good agreement with a recent study performed by Groß et al. (2014b); taken together, these two studies show that the rate of OH regeneration through Reaction (R5) is more rapid than previously thought. GEOS-Chem has been used to assess the implications of the revised rate coefficients and branching ratios; the modelling shows an enhancement of up to 5 % in OH concentrations in tropical rainforest areas and increases of up to 10 % at altitudes of 6–8 km above the equator, compared to calculations based on the IUPAC recommended rate coefficient and yield. The enhanced rate of acetylperoxy consumption significantly reduces PAN in remote regions (up to 30 %) with commensurate reductions in background NOx .