Chronic bronchitis, defined as increased cough and sputum production, is well established as an occupational disease independent of cigarette smoking, which is a major confounding factor. The relevance of industrial exposure to the development of airflow obstruction (chronic obstructive pulmonary disease (COPD)) is more controversial. Good data exist linking the inhalation of cadmium fume to the development of emphysema. The pathology of airflow obstruction due to other occupational exposures is less clear. The principal measure of occupational bronchitis with airflow obstruction is accelerated loss of forced expiratory volume in one second (FEV1); and epidemiology is the principal tool for investigation. Using these methods, many workers have established links between occupational exposure and disease, which will be reviewed. From an occupational standpoint tobacco smoking is unique, in that the pollutant is put into the mouth and deliberately inhaled in concentrations far exceeding occupational exposures. Despite this, less than 20% of smokers develop significant airflow obstruction. The factors which determine whether an individual smoker develops airflow obstruction are usually unknown. There are two ways in which an occupational agent may act: 1) by promoting the deleterious effects of smoking; or 2) by acting in a manner similar to tobacco smoking, requiring other promoting factors before an effect is seen. Examples will be given of both of these mechanisms. From a pragmatic point of view, it is highly unlikely that the mixed agents in tobacco smoke are the only cause of significant COPD. The widespread habit of tobacco smoking in industrial populations has delayed the recognition of other factors contributing to disease, as it is only possible to examine the independent risks of smoking and occupational exposures when a reasonable proportion of those occupationally exposed are lifelong nonsmokers. The studies of specific workforces are often difficult to interpret, as workers commencing exposure in jobs with a high risk of lung disease have even better lung function than the normal worker, and losses from a cohort are higher in those with disease, leading to lower estimates of risk as an increasingly smaller proportion of survivors are followed-up. These problems have been overcome by sampling the general population away from the workplace. Such studies suffer from less selection bias (but exclude those who have died prematurely), and have less satisfactory estimates of occupational exposure, usually including a greater proportion of those with low exposure and, thus, reducing the relative risks of exposure. Such studies have included a general older working population in Beijing (China) [1], an urban and rural population in Norway [2], a rural population in Northern Italy [3], and two studies of city dwellers designed to investigate air pollution in France [4] and the USA [5]. Different outcome measures have been used, but all show significant excess risks with occupation. In general, dust exposure is associated with higher risks of reduced lung function than gas or fume exposure [1, 5], with relative risks of around 1.5. General population samples have been studied longitudinally in Paris [6], Cracow (Poland) [7], Zutphen (Holland) [8] and Bergen (Norway) [9]. The Zutphen study again started with a community sample, using a job matrix to link symptoms and diagnoses with the occupation. The study has been analyzed longitudinally, initially using the time, intensity and duration of exposure. The longitudinal analysis found a cumulative 25 yr incidence of chronic nonspecific lung disease of 27.9%. Incidence density ratios were calculated, and ratios for smoking increased from 1.5 for light smokers to 4.5 for those smoking more than 20 cigarettes daily. Increased risks of chronic nonspecific lung disease (which includes asthma), were found in a number of occupations which are shown in table 1. A further analysis of this study, by the same group, appears in this issue of the Journal [10], using a different job exposure matrix generated from the sample itself. It is disappointing that the results of the reanalysis are less clear than the original analysis, perhaps because of the lack of differentiation between the proportion of the workforce exposed to a particular agent, and its intensity. The mortality analysis was limited by the small number of deaths (33 out of 799) due to chronic nonspecific lung disease. Nevertheless, for occupational groups where at