Primary malignant mesothelial tumours were recognized by pathologists before asbestiform minerals (chrysotile, crocidolite and amosite) were mined commercially. The discovery, 40 yrs ago, of a causal link with crocidolite and the wide-ranging epidemiological studies which followed are the subject of this review. Early case-control and descriptive surveys, supplemented by cohort studies in insulation workers and chrysotile miners, quickly demonstrated major occupational and geographical differences, with high risk in naval dockyard areas and in the heating trades. In the 1980s, reliable cohort surveys showed that in mining and in the manufacture of asbestos products the mesothelioma risk was much higher when exposure included crocidolite or amosite than chrysotile alone. However, qualitative and quantitative information on exposure was too often inadequate for this evidence to be conclusive. Well-controlled lung fibre analyses have reduced these deficiencies and demonstrated the probable implications of the greater biopersistence of amphibole fibres. Chrysotile for industrial use often contains low concentrations of fibrous tremolite, which may well explain the few cases of mesothelioma associated with this type of asbestos. Progress in this field has been much retarded by controversy, for which the 20 year gap between the availability of reliable estimates of risk for the mining of chrysotile and that for crocidolite or amosite may have been largely responsible.