Context The French cohort of uranium miners was set up in the 1980’s to study the risk of mortality associated with occupational ionising radiation (IR) exposure. The analyses focused on radon exposure for a long time, but the exposure to long-lived radionuclides (LLR) arising from uranium ore dust and to external gamma rays is now also considered in the latest analyses. The individual assessment of IR exposure improved over time. The aim is to present an overview of these strengthening in the frame of the estimation of lung cancer risk.

Methods The French cohort includes 5,086 men employed as a miner for at least one year and followed-up for 35 years in average (1946–2007). At end of follow-up, 211 cases of lung cancer deaths were identified among the 1935 deaths. Measurement errors (ME) in radon exposure assessment were characterised according to the periods and the methods of exposure measurement (retrospective, ambient, individual assessment). Gamma rays and LLR exposures were monitored from 1956, allowing the calculation of the absorbed lung dose due to the three radiation components.

Results The size of ME in radon exposure decreased from 94% in the earliest years to 10% in the recent years. In the absence of correction for ME, the assessment of the risk for lung cancer death could be attenuated by up to 60%. In terms of lung doses, radon gas and progenies contributed to 58% of the total absorbed dose and to 97% of the equivalent dose, supporting the association between lung cancer risk and radon exposure.

Conclusion Currently, a new approach is underway to take ME into account by using a structural Bayesien method. Furthermore, an international collaborative analysis of active miner cohorts would improve the estimation of risks for cancer or non-cancer diseases.