Ozone, 03, reacts with a carbon sample at room temperature. Clean carbon samples show a half to one and a half order of magnitude increased initial rate constant (k,) for O3 loss relative to repetitively exposed carbon samples. The ozone loss rate and therefore the rate constant reaches steady state (k8J on the time scale of tens of minutes, upon exposure to a characteristic dose of 8 x 1017 molecules for a 30-mg carbon sample independent of the flow rate. This characteristic dose closely corresponds to a monolayer of adsorbed ozone molecules on the carbon sample. Both k, and k,. decrease with increasing flow rate of O3 into the reactor, and the loss rate is found to depend on Lo3]. When the loss rate is plotted against the steady state concentration of 0 3 , a saturation plot results which is proportional to the surface coverage, 8, at a given [03]. This interpretation rests upon a Langmuir type kinetics model with an assumed first-order dependence of the loss rate constant. The "sticking coefficients" track the rate constants and are on the order of to depending on the carbon sample, dose, and flow rate. Furthermore, k , depends on the length of the dark period (absence of O3 exposure) and is larger the longer time the sample has had to recuperate from previous O3 exposures up to a period of 150 s. This surface relaxation is thought of as a time-dependent change in surface coverage taking place in the dark period and is therefore an indication of a slow surface diffusionireaction that can be separated from the adsorption-desorption kinetics. The mass balance shows that for every ozone molecule that is lost on the surface, an oxygen molecule is found. This adsorbed odd oxygen is then responsible for product formation, which comprises the volatile components CO and CO, to an extent of 20 to 40% of the odd oxygen deposited on the surface. The difference is thought to be in a preoxidized state on the carbon sample, which evolves CO and GO, upon heat treatment.