Atmospheric hydroperoxides (ROOH) were measured at Summit, Greenland (72.97° N, 38.77° W) in summer 2003 (SUM03) and spring 2004 (SUM04) and South Pole in December 2003 (SP03). The two dominant hydroperoxides were H<sub>2</sub>O<sub>2</sub> and CH<sub>3</sub>OOH (from here on MHP) with average(±1σ) mixing ratios of 1448(±688) pptv, 204(±162) and 278(±67) for H<sub>2</sub>O<sub>2</sub> and 578(±377) pptv, 139(±101) pptv and 138(±89) pptv for MHP, respectively. In early spring, MHP dominated the ROOH budget and showed night time maxima and daytime minima, out of phase with the diurnal cycle of H<sub>2</sub>O<sub>2</sub>, suggesting that the organic peroxide is controlled by photochemistry, while H<sub>2</sub>O<sub>2</sub> is largely influenced by temperature driven exchange between the atmosphere and snow. Highly constrained photochemical box model runs yielded median ratios between modeled and observed MHP of 52%, 148% and 3% for SUM03, SUM04 and SP03, respectively. At Summit firn air measurements and model calculations suggest a daytime sink of MHP in the upper snow pack, which decreases in strength through the spring season into the summer. Up to 50% of the estimated sink rates of 1–5×10<sup>11</sup> molecules m<sup>−3</sup> s<sup>−1</sup> equivalent to 24–96 pptv h<sup>−1</sup> can be explained by photolysis and reaction with the OH radical in firn air and in the quasi-liquid layer on snow grains. Rapid processing of MHP in surface snow is expected to contribute significantly to a photochemical snow pack source of formaldehyde (CH<sub>2</sub>O). Conversely, summer levels of MHP at South Pole are inconsistent with the prevailing high NO concentrations, and cannot be explained currently by known photochemical precursors or transport, thus suggesting a missing source. Simultaneous measurements of H<sub>2</sub>O<sub>2</sub>, MHP and CH<sub>2</sub>O allow to constrain the NO background today and potentially also in the past using ice cores, although it seems less likely that MHP is preserved in firn and ice.