The reaction probabilities, &ggr;, ClONO2 and N2O5 on ice and nitric acid trihydrate (NAT) surface were determined, using reactant concentrations that are typical of the lower stratosphere, by measuring the first-order reactant loss rate coefficients over the substrate located on the walls of a fast-flow reactor. Reactants were detected using chemical ionization mass spectrometry, a sensitive technique which allows the use of low reactant concentrations. The reaction probabilities obtained for ClONO2 are: 0.3 (+0.7-0.1) on pure ice, 0.006 (¿30%) over a NAT surface, and 0.3 (+0.7-0.1) on an HCl-doped NAT surface. Those for N2O5 are 0.024 (¿30%) on pure ice, 0.0006 (¿50%) on a NAT surface, and 0.0032 (¿30%) on an HCl-doped NAT surface. We observed that an ice surface will be converted into a less reactive HNO3-doped ice surface in a relative short time, and we present arguments that this surface consisted of a NAT layer. The large differences between our results and previous measurements for some of these &ggr; can be attributed to the relatively large reactant concentrations used in the previous work. The major implications of this work for polar stratospheric chlorine activation are: an efficient loss of ClONO2 on pure ice surfaces, a very rapid rate for the reaction ClONO2+HCl on NAT surfaces, and that pure ice surfaces will become ''passivated'' when coated with one monolayer of NAT crystal. ¿ American Geophysical Union 1991