The increasing number of motor vehicles is associated with diffuse exposure to their exhausts, formed by mixtures of hundreds of chemical compounds. Some of these compounds cause chronic irritation of the respiratory mucosa, while others are carcinogenic in experimental animals, particularly polycyclic aromatic hydrocarbons and nitroarenes. This review aims to describe epidemiologic evidences of carcinogenicity of engine exhausts. The study of the effects in humans of the exposure at issue is impaired by several difficulties: widespread exposure of a large part of the population, with low levels of exposure; aspecific effects, shared with other exposures; concurrent exposures acting on the same targets, particularly the respiratory mucosa (smoking, occupational exposures); difficulties in the quantitation of exposure to exhaust. We review the epidemiologic studies concerning occupational exposures published up to June, 1988. Overall, excess risks for lung and bladder cancers and for hematolymphopoietic malignancies have been reported, particularly in people exposed to diesel exhaust. Excesses are modest, and their interpretation is often impaired by lacking or inadequate consideration of the effects of potential confounders. The review considers, for each study, enrollment procedures, information on exposure, and inclusion of potential confounders. In addition, aspects concerning internal coherence, such as a dose-response relationship, are considered. Particularly relevant are two studies on USA railroad workers, showing a relative risk of 1.4 among those exposed to diesel exhaust. The authors used, in the estimation of exposure, actual measurement of dose in recent years for some jobs and workplaces. Levels of exposure to respirable dusts were around 82-330 micrograms/m3, after allowing for cigarette smoking. According to the International Agency for Research on Cancer, these is limited evidence of carcinogenicity in humans of diesel exhaust, and inadequate evidence for gasoline exhaust.