Over the last 25 years, railroad workers and truck drivers occupationally probably exposed to diesel exhaust were investigated in numerous epidemiological studies for prevalence of lung cancer. A recent assessment of 21 of these studies, including 11 truck driver studies by the Health Effects Institute (HEI, 1995), claimed that the studies showed consistently a lung cancer risk increase of 20 to 40% for the study subjects. The supposedly most reliable study of railroad workers in the United States is presently under serious dispute. With regard to the assessed truck driver studies, uncontemplated, incomplete, or incorrect reporting on more than half of the 11 truck driver investigations warranted a reanalysis of the original study results. The quality and power of the truck driver studies varied widely. Two older studies - one of them showing significant risk increases - were already rejected in 1981 by the National Research Council in the United States as being inadequate. Of 7 studies reporting nonsignificant results, 6 indicated increases and 1 showed a decrease of lung cancer risk. In 4 studies, no adjustment was made for the smoking habits of the study groups and their controls. The reported risk increases - two of them supposedly significant - could be due to the lack of smoking adjustment. Most of the studies compared the truck drivers to the general population, although the unusual working conditions of long-haul truck drivers effected an unhealthy way of life along the road that could easily be a stronger confounder in comparison to the general public than cigarette smoking. A study supposedly showing a significant relative risk of 2.3 was particularly poor with regard to correction of smoking habits and lack of information on potential exposure. In another study, the original investigators aggravated their results by using one-sided p-tests and 90% confidence intervals. At least two of the HEI analyses were flawed. One of the two better qualified hospital-based case-control studies shows a maximum likelihood odds ratio below 1.0. The HEI report, however, indicated an erroneous value in excess of 1.0, which incorrectly supports the HEI conclusion that the observed lung cancer increases of 20 to 40% are consistent. The other qualified study was an earlier, equivalent investigation where adjustment for cigarette smoking was incomplete and where lifestyle confounding was not considered. Despite these drawbacks, the adjusted data shifted the unadjusted risk from significant increases in the crude analysis to statistically nonsignificant risk elevations after controlling for smoking. The authors of this reanalysis concur with the position taken by the HEI Board of Directors that "the average levels of diesel exhaust found in most occupational settings, which are below 100 mug/m 3, would not likely be a cancer hazard for these workers, nor would ambient levels (1 to 10 mug/m 3) present a cancer risk for the general population" (HEI, 1995, p.1). Moreover, the critical reanalysis of the truck driver studies in the background paper of the HEI document leads to the conclusion that the proposed general assessment of a small but consistent increase of lung cancer risk in possibly diesel-exposed truck drivers is unfounded and is an overinterpretation of systematic errors due to neglected or incomplete control of confounding by excessive smoking, as well as by the irregular lifestyle of the truck drivers in comparison to the control population. Deplorably, and unlike experimental studies acceptable for regulatory decisions, there are no qualification standards for epidemiological studies, and data have been used by regulators to establish graded degrees of evidence even in view of disclaimers by the original authors. Thus, the poorest hypotheses generating pilot studies, such as the ones rejected by the National Research Council in 1981, had a chance to participate in the weighing of evidence in the HEI background paper 14 years later.