The rate and thermodynamics of the adsorption of acetone on ice surfaces have been studied in the temperature range T = 190-220 K using a coated-wall flow tube reactor (CWFT) coupled with QMS detection. Ice films of 75 +/- 25 mu m thickness were prepared by coating the reactor using a calibrated flow of water vapor. The rate coefficients for adsorption and desorption as well as adsorption isotherms have been derived from temporal profiles of the gas phase concentration at the exit of the flow reactor together with a kinetic model that has recently been developed in our group to simulate reversible adsorption in CWFTs (Behr, P.; Terziyski, A.; Zellner, R. Z. Phys. Chem. 2004, 218, 1307-1327). It is found that acetone adsorption is entirely reversible; the adsorption capacity, however, depends on temperature and decreases with the age of the ice film. The aging effect is most pronounced at low acetone gas-phase concentrations (<= 2.0 x 10(11) molecules/cm(3)) and at low temperatures. Under these conditions, acetone is initially adsorbed with a high rate and high surface coverage that, upon aging, both become lower. This effect is explained by the existence of initially two adsorption sites (1) and (2), which differ in nature and number density and for which the relative fractions change with time. Using two-site dynamic modeling, the rate coefficients for adsorption (k(ads)) and desorption (k(des)) as well as the Langmuir constant (K-L) and the maximum number of adsorption sites (C-s,C-max), as obtained for the adsorption of acetone on sites of types (1) and (2) in the respective temperature range, are kads(1) = 3.8 x 10-(14) T-0.5 cm(3) s(-1), k(des)(1) = 4.0 x 10(11) exp(-5773/T) s(-1), K-L (1) = 6.3 x 10(-25) exp(5893/T) cm(3), C-s,C-max(1) <= 10(14) cm(-2) and k(ads)(2) = 2.9 x 10(-15) T-0.5 cm(3) s(-1), k(des)(2) = 1.5 x 10(7) exp(-3488/T) s(-1), K-L(2) = 5.0 x 10(-22) exp(3849/T) cm(3), C-s,C-max(2) = 6.0 x 10(14) cm(-2), respectively. On the basis of these results, the adsorption of acetone on aged ice occurs exclusively on sites of type (2). Among the possible explanations for the time-dependent two-site adsorption behavior, i.e., crystallographic differences, molecular or engraved microstructures, or a mixture of the two, we tentatively accept the former, i.e., that the two adsorption sites correspond to cubic (1, I-c) and hexagonal (2, I-h) sites. The temporal change of I-c to I-h and, hence, the time constants of aging are consistent with independent information in the literature on these phase changes.