Due to the absence of sunlight, unexpected high nighttime OH concentrations reported in previous field studies are of high interest for in-depth understanding of trace gas removal and reaction kinetics. In summer 2006, within the framework of PRIDE-PRD2006 and CAREBEIJING2006, we performed intensive in-situ measurements for HOX radicals and ancillary parameters at two non-urban sites in Pearl River Delta and Beijing, respectively. During nighttime, similar features for both campaigns were observed. Radical concentration decreased in the afternoon to values of 3×106cm?3 (OH) and 5×108cm?3 (HO2) of sunset. After sunset, concentration decreased steadily even further reaching 0. 5×106cm?3 (OH) and 0. 2×108cm?3 (HO2) shortly before sunrise. Consequently, the averaged nighttime pollutant turnover rates by OH were as high as 8 ppb/h and 4 ppb/h for PRD and Beijing, respectively, dominating nighttime oxidation processes. A box model with the established chemical mechanism (RACM-MIM-GK) underestimated these observed OH concentrations by an order of magnitude while reproduced the observed HO2 concentrations taking into account the known interference from ambient RO2 radicals. By testing the recently proposed recycling mechanisms applied for daytime chemistry, we found both a small primary source and a secondary source of OH radicals were required additionally, of which the last one comparable to daytime observation. Interestingly, the widely applied Leuven Isoprene Mechanism showed marginal impacts on the modeled nighttime OH concentrations under high isoprene and low NO conditions. With the help of a simple one dimensional simulation, we found that direct input of ROx radicals by vertical transport was negligible while the input of PAN and MPAN could be of significance.