A mortality study was carried out in conjunction with the European mortality study among welders coordinated by the International Agency for Research on Cancer (IARC). The study was aimed at assessing risks for lung cancer in relation to exposure to asbestos, welding fumes containing chromium and nickel, and tobacco smoke. The study included a cohort of 2721 welders and an internal comparison group of 6683 manual workers employed in 13 factories in France. The mortality of the two cohorts was studied from 1975 to 1988 by the historical prospective method. Job histories of welders were traced including welding processes used, metals welded, and proportion of worktime spent in welding. Data on smoking habits were collected from medical records. The observed number of deaths were compared with those expected (standardised mortality ratio (SMR)) based on national rates with adjustments for age, sex, and calendar time. The smoking habits of 87% of the whole study population were known. The distribution of welders and controls according to smoking was not statistically different. The overall mortality was slightly higher for welders (SMR = 1.02, 95% confidence interval (95% CI) 0.89-1.18) than for controls (SMR = 0.91, 95% CI 0.84-0.99). For lung cancer, the SMR was 1.24 (95% CI 0.75-1.94) for welders, whereas the corresponding value was lower for controls (SMR = 0.94, 95% CI 0.68-1.26). The SMR for lung cancer was 1.59 among non-shipyard mild steel welders (95% CI 0.73-3.02). This contrasted with the results for all stainless steel welders (SMR = 0.92, 95% CI 0.19-2.69), and for stainless steel welders predominantly exposed to chromium VI (SMR = 1.03, 95% CI 0.12-3.71). Moreover, SMRs for lung cancer for mild steel welders tended to increase with duration of exposure and time since first exposure, leading to significant excesses for duration > or = 20 years and latency > or = 20 years. Such a pattern was not found for stainless steel welders.