Background Using data from a Dutch cohort of workers in road construction and asphalt mixing companies, this article describes possible confounding of the association between exposure to bitumen fume and lung cancer mortality by smoking. Methods A retrospective cohort of 3,714 workers with at least one season of employment was identified. Semi‐quantitative exposure to bitumen fume was assessed by a job‐exposure matrix. Information on smoking habits was available for a sub‐cohort of 1,138 workers, who underwent medical examinations by the occupational health services in the past. Results Smoking habits differed between occupational title groups and there was a positive association between cumulative exposure and smoking. Internal analyses using the non‐exposed subjects as reference category, showed a positive association between semi‐quantitative bitumen fume exposure and lung cancer risk. After adjusting for differences in smoking habits, all relative risks were reduced, but a weak positive association could still be observed. Conclusion Confounding by smoking on the association between exposure to bitumen fume and lung cancer mortality is possible, although the positive trend (not statistically significant) for lung cancer mortality remained. Only a nested case‐control study may allow proper treatment of potential (residual) confounding by smoking in this population. Am. J. Ind. Med. 43:79–87, 2003. © 2003 Wiley‐Liss, Inc.