To understand the atmospheric stability of levoglucosan, which is a major molecular tracer used for source apportionment of biomass burning aerosols, degradation kinetics of levoglucosan by hydroxyl radical (OH) have been investigated using a flow reactor under different conditions. The second-order rate constant (k(2)) for the degradation of pure levoglucosan by OH is (9.17 +/- 1.16) x 10(-12) cm(3) molecules(-1) s(-1) at 25 degrees C and 40% relative humidity (RH), while it depends on environmental conditions such as temperature, RH, and mixing state. At 25 degrees C, k2 of pure levoglucosan linearly decreases with increasing RH (k(2) = (1.50 +/- 0.04) x 10(-11) - (1.31 +/- 0.11) x 10(-11)RH), while it increases with increasing temperature and follows the Arrhenius equation k(2) = (6.2 +/- 5.6) x 10(-9)exp[(-1922.5 +/- 268.2)/T] when the RH is 40%. At 25 degrees C and 40% RH, compared to pure levoglucosan, levoglucosan coated on (NH4)(2)SO4 or NaCl (levoglucosan@(NH4)(2)SO4 and levoglucosan@NaCI) shows larger k(2) to OH with (9.53 +/- 0.39) x 10(-12) and (10.3 +/- 0.45) x 10(-12) cm(3) molecules respectively, whereas levoglucosan coated on soot (levoglucosan@soot) shows the smaller k(2) of (4.04 +/- 0.29) x 10(-12) cm(3) molecules(-1) s(-1). Either (NH4)(2)SO4 or NaCI internally mixed with levoglucosan ((NH4)(2)SO4@levoglucosan and NaCl@levoglucosan) prominently inhibits the degradation of levoglucosan. Based on the rate constants, atmospheric lifetimes of levoglucosan were estimated to be 1.2-3.9 days under different conditions. All the results indicate that the degradation of levoglucosan by OH is prominent during air mass aging, and it should have an important influence on the uncertainty of source apportionment if the contribution of degradation to levoglucosan concentration is not considered in source apportionment models. (C) 2014 Elsevier Ltd. All rights reserved.