The chemical amplifier (CA) was used to measure the concentration of the peroxy radicals present in the nighttime atmosphere at the Mauna Loa Observatory (MLO) during the Mauna Loa Observatory Photochemistry Experiment (MLOPEX 2) from April 15 to May 15, 1992. Simultaneous measurements were also made of peroxy radicals using a modified OH instrument during one period of 56-hour duration. Both instruments observe experimentally significant nighttime signals, but the magnitude of the larger CA signal appears to result from organic peroxy radicals to which the modified OH instrument is insensitive. The data from about one half of the nights showed a decay of the radical signal from 2000 to 0400 HST the next morning. The decay rates followed roughly those expected for primary or secondary alkyl peroxy radicals. However, data from the other half of nights showed an increase in radical concentration with time, which reflected significant radical generation rates. Several alternatives are considered in explanation of the results. (1) The possible origin of the signals from the interference of HO2NO2, presumably transported from reservoirs at higher elevations, was considered. The data are inconclusive, and this hypothesis is tentatively rejected. (2) The nighttime chemistry of the O3 and NO3 was also considered. The observed trace gases provide in theory only a small fraction of the peroxy radical signal seen on many nights. However, the presence of 1--30 parts per trillion by volume (pptv) of CH3SCH3, a compound not measured but expected to be present at MLO, can react with NO3 to produce peroxy radical signals of the magnitude observed. It is concluded that this ubiquitous compound over the oceans offers the best current hypothesis to rationalize the nighttime generation of peroxy radicals at MLO.¿ 1997 American Geophysical Union