Studies of workers have played a central role in identifying human carcinogens. For more than a third of the approximately 100 agents classified as carcinogenic (Group 1) by IARC with sufficient evidence in humans, the critical evidence was provided by occupational epidemiologic studies. Data from occupational studies has also contributed important evidence to identifying over 300 possibly or probably carcinogenic agents. Nevertheless, the contribution of occupational epidemiology could be improved. Beyond clear reporting of methods and results, the greatest need is for quantitative assessment of exposure and analysis of exposure-response relations. The political context of carcinogen assessment imposes barriers of a different kind. Actions by actors with vested interests to intimidate scientists, stifle debate and derail risk assessment are well documented. Recent evaluations of the herbicide glyphosate will be discussed as a case example.