Soot samples from a spark generator, a flame, and a diesel passenger car were either collected on a Teflon filter and transferred to an IR-transparent window or deposited directly from a flame onto the window and investigated by Fourier transform infrared (FTIR) spectroscopy. The soot-covered windows were mounted in a 10 cm vacuum cell connected to a standard flow system with He as carrier gas. Reactive gases, such as NO2 and HNO3, were added to the carrier gas flow at a concentration of (0.016 to 2.5) x 10(14) molecule cm(-3). FTIR spectra of soot samples before and after exposure to HNO3, NO2, and O-3 are presented. Formation of IR absorption bands was analyzed as a function of exposure time. IR bands attributable to soot surface oxidation products and nitrogen containing species, e.g. -C=O, R-NO2, R-ONO2, and R-ONO were observed. The observed time dependence of the absorption bands of the spark generator soot can be fitted by two parallel reactions, a slow and a fast process. Both processes have a reaction order of n approximate to 0.2 (+/-0.3) for the NO2 + soot reaction and n approximate to 0.5 (+/-0.6) for the HNO3 + soot reaction. The number of active sites, N-max = 2.2 x 10(14) molecules cm(-2) soot surface, has been estimated from saturation experiments. Surface reaction probabilities depend on reactant concentration and reaction time and were in the range of gamma approximate to 10(-6) to 10(-8) for the slow, and gamma approximate to 10(-3) to 10(-6) for the fast processes. The reaction probability on diesel engine soot was nearly 1 order of magnitude slower. It is concluded that the reaction of NO2, with soot cannot account for the HONO levels observed in urban air.