The genotoxic activity of benzo[a]pyrene (BAP), 2-nitrofluorene (NF) and airborne particulate matter was evaluated in the DNA-repair host-mediated assay after intraperitoneal or intratracheal administration. Dimethylnitrosamine (DMNA), used as a positive control, showed a genotoxic effect after both intraperitoneal and intratracheal administration, the strongest effect being found in liver, followed by lungs and kidneys, whereas a weak effect was observed in the spleen. In general no difference in genotoxicity was found between the 2 administration routes used. For BAP, although clearly positive in vitro, a moderate dose-dependent effect was found only in the liver after intraperitoneal administration. NF, which was positive in vitro both with and without a metabolizing system, produced no genotoxic effect in any of the organs tested after intraperitoneal administration. Extracts of airborne particulate matter which were genotoxic in vitro failed to cause a genotoxic effect in vivo by either route of administration. Possible explanations for the differences between the data obtained in vitro and in vivo are discussed.