A follow-up study over 12 years was conducted among 556 men aged 30 to 54 in 1960 and working at that time in factories around Paris (France). Various occupational exposures were recorded at the time of the 1960 survey after a technical study of each workplace. The annual rate of decline of FEV1 during 12 years was estimated for each subject from the measurements in 1960 and 1972. This rate (the FEV1 slope) was related independently of FEV1 level (which reflects the loss since the beginning of adult life) and of smoking habits to occupational exposure to dust, gases, and heat. FEV1 slope was significantly related to inhalation of mineral dust (even in the absence of silica) as well as to grain dust, and the slope was steeper with increased intensity of exposure to dust. Analysis of job changes showed that among heavily exposed subjects, those who changed jobs had a less steep slope than those who did not. Our results support the hypothesis of a causal role of exposure to dust in the development of chronic airflow obstruction and of a benefit when exposure to dust ceases. Exposure to dust, gas, and heat usually occurred together so data on gas and heat were analysed after taking account of exposure to dust. The influence of heat on FEV1 decline showed a clear trend. Results suggest that exposure to gases associated with exposure to dust or heat or both had a deleterious effect. After adjusting for age, smoking, and FEV1 level (ASLA) the following average slopes were obtained: 44 ml/a (for exposure to none or to only a slight amount of dust, or to gases alone), 51 ml/a (heat), 53 ml/a (noticeable dust), 55 ml/a (noticeable dust and heat), 60 ml/a (noticeable dust, heat, and high concentration of gases). Independently of the occupational exposures, ASLA FEV1 slopes among manual workers were related to skill, being 44 ml/a for skilled and 51 ml/a for unskilled men. Independently of social class and occupational exposures recorded, there were differences in FEV1 slopes by factory, suggesting that one should not rely on using one factory as the control of studies of occupational exposure to another.