Résumé : The exposure of asbestos was studied in 31 cases of mesothelioma from case histories and by microscopic mineralogical analysis of the broncho-alveolar lavage fluid (BAL) (31 cases) and of pulmonary tissue (5 cases). When definite exposure had occurred the lavage contained asbestos bodies in every case, except one patient with severe airflow obstruction. The most positive lavages, probably reflecting exposure to industrial amphiboles, were found in patients presenting with radiological evidence of asbestosis. Two patients had positive lavages 50 years after exposure had ceased. The lavage confirmed exposure in 6 out of 11 suspected cases and revealed contamination by asbestos in a further 4 cases, in a group of 6 not known to have been exposed. In 19.3% of cases there was a low concentration of asbestos bodies (less than 1AB/ml of BAL), comparable to what is found in 16.2% of controls from an urban population. An analysis of lung tissue confirmed massive exposure (greater than 20 000 AB/g) in two cases and in particular revealed a significant dust load (greater than 250 AB/g) in patients presenting with a weak positive BAL. Thus it seems that all the positive results should be taken into account. Of the 31 cases, 8 had a BAL without asbestos bodies. These were either mesotheliomas not linked to the inhalation of asbestos, or were the result of false negatives on account of artefacts related either to the BAL technique itself or linked to the technique of mineral analysis. Indeed the microscopic counts of asbestos bodies probably underestimate certain exposures, notably environmental exposure to chrysotile.