This paper reviews animal evidence of carcinogenicity, discusses problems in assessing exposure to diesel exhaust in human epidemiologic studies, reviews human epidemiology on lung cancer, and discusses ongoing epidemiologic research. Animal studies on carcinogenicity of diesel exhaust via inhalation have yielded differing results. Human epidemiologic results were reported for miners, truck drivers, bus company employees, railroad workers, and heavy equipment operators. A major difficulty in human occupational studies has been measurement of actual diesel exposure, since there is no standard method for such measurements. Diesel exhaust is known to contain carcinogenic compounds which are attached to respirable particulate, and there are animal data to indicate that diesel exhaust can act as a lung carcinogen. Epidemiologic studies have been inconclusive, partly because of difficulty in accurately assessing historical exposures. Since the introduction of diesel engines into a number of occupations is relatively recent, insufficient potential latency is also a factor. Epidemiology of lung cancer and diesel exhaust is particularly difficult because gasoline exhaust, cigarette smoke, and general urban air pollution contain many of the same suspect compounds which are contained in diesel exhaust. The authors conclude that if diesel exhaust is a human lung carcinogen, it appears to be a weak one. Hence, control over smoking, a major confounder, is especially important. It is difficult to control for smoking in retrospective occupational cohort studies, although such a study design has important advantages over population based case control studies. Case-control studies nested within cohorts exposed in the past may provide a better approach, especially if there is a group nonexposed or minimally exposed within the cohort.