Incidence rates for non-Hodgkin's lymphoma (NHL) have been rising throughout the world for several decades, and no convincing explanation exists for the majority of this increase. The commonest subtypes of NHL have no well-defined aetiological factors but lymphoma development has been linked with exposure to a variety of chemicals, including nitrates, pesticides, herbicides, and solvents. Benzene, a solvent and important constituent of petrochemical products, is a potent lymphomagen in experimental animals and high-dose exposure in humans is associated with both acute myeloid leukaemia and NHL. Much current interest centres on the possibility that environmental benzene exposure in the general public may underlie a proportion of the increase in NHL. Seventy per cent of benzene exposure in the environment is derived from vehicle exhaust emissions, whose increase has closely paralleled the rise in frequency of the disease. Mathematical modelling has been used to calculate an acceptable concentration of benzene in air based on risk estimates derived from industrial exposure, but the recommended target concentration in the U.K. of 1 ppb is regularly exceeded in urban locations. Detailed investigation of the health effects of low-level benzene exposure awaits an accurate assay for quantifying long-term human exposure. The 32 P post-labelling technique for the detection of toxin-specific DNA adducts is extremely sensitive and has been applied in the biomonitoring of exposure to a number of carcinogens, but benzene-DNA adducts have to date proved elusive of detection.