An epidemiological, cross-sectional study was conducted in order to assess non-neoplasic effects on the lung due to chronic exposure to arc welding fumes and gases. The study involved 346 arc welders and 214 control workers from a factory producing industrial vehicles. These workers (welders and controls) had never been exposed to asbestos. Respiratory impairments were evaluated by using a standardized questionnaire, a clinical examination, chest radiophotography and several lung function tests (spirometry, bronchial challenge test to acetylcholine, CO transfer tests according to the breath-holding and the steady-state methods, N2 washout test). The only significant differences between the welders overall compared to the controls were a slightly higher bronchial hyper-reactivity to acetylcholine and a lower lung diffusing capacity for CO in the welders. However, non-specific, radiologic abnormalities (reticulation, micronodulation) and obstructive signs were more frequent in the most exposed welders (welding inside tanks) than in welders working in well ventilated workplaces. The nature of the metal welded (mild-steel, stainless steel, aluminium) did not seem to have an influence on respiratory impairments. In the mild-steel welders, respiratory symptoms (dyspnoea, recurrent bronchitis) and obstructive signs were more frequent in the welders using a manual process than in the welders involved with the semi-automatic process (MIG). For all the workers (welders and controls), smoking had a markedly adverse effect on respiratory symptoms and lung function. Moreover, smoking seemed to interact with welding since CO lung transfer was more impaired in smoking welders than in smoking controls.