From an original prospective cohort of 244 current and ex-workers in two asbestos cement plants, longitudinal radiographic data covering ten years were available for 165 and lung function data covering about six years for 150. Estimates of average and cumulative dust exposure were available for each participant, all men. Radiographic progression (onset or worsening) was assessed by comparing earliest and latest films side by side. Annual changes in lung function were computed by fitting regression lines to all the data points. Small opacities (ILO category 1/0 or higher) were found in 16% of initial films, and progression of small opacities occurred in 13% of film pairs. Average and cumulative dust exposure were each significant determinants of the initial presence of small opacities, and were determinants of the progression of both parenchymal and pleural abnormalities. There was greater likelihood of progression if an abnormality was initially present, and a greater likelihood of progression in the plant that had systematic use of some crocidolite. Initial levels of lung function were related to smoking, exposure to dust, and initial radiographic status. Mean annual declines in lung function were modest (FVC-0.017 l/y, FEV1-0.020 l/y) and were related to smoking but not exposure to dust, initial radiographic status, or radiographic progression. Both plants used mainly chrysotile asbestos and exposure levels declined severalfold after 1960. Our findings suggest a waning effect of the larger remote dust exposures on recent annual change in lung function. This accords with human and experimental pathology data showing the relatively low resistance of chrysotile fibres to chemical alteration and clearance.