Urban air pollution, originating in western countries mainly from automotive engine exhausts, contains thousands of components, many of which are genotoxic, i.e. are putative cancer initiators. Other pollution components, such as NO2and certain particles, may have cocarcinogenic/promotive effects, at least at higher exposure levels. Cancer risk assessment of this complex mixture has to combine data from the exposure history, from epidemiological studies as well as from animal carcinogenicity tests, and from in vitro studies of fractions and individual components. Data for metabolism and pharmaco-kinetics have also to be considered. A multiplicative linear model is assumed to be valid for cancer initiation at low levels. Attempts are being made to determine the target dose from ultimate carcinogens (reactive metabolites)viamacromolecule adduct levels, and to base the risk assessment on the radiation-dose equivalent to the chemical dose. So far this has been possible only for simple alkenes, which are metabolized to epoxides, and indirectly, via benzo(a)pyrene (BaP), for particle-bound polycyclic aromatic hydrocarbons (PAH). The lifetime risk of cancer (all sites) from ethene is estimated accordingly to 1.4 x 10-4per fig µg m-3, and from PAH to 12 x 10-4per ng m-3BaP. For other components indicated to give risk contribution (NOx, volatile PAH, benzene, aldehydes, butadiene) essential data are lacking and only very rough estimates can be given at this time.