Résumé : Asbestos body (AB) countings on both bronchoalveolar lavage (BAL) fluids and digested lung tissue samples were performed in one hundred consecutive subjects submitted to a thoracotomy procedure, mostly for lung carcinoma. A good correlation (r = 0.73) was found between the two groups of values for the total group of subjects. When restrictive selection criteria were taken into account, such as lavage homolateral to the analysed lung, performed by the same trained physician, this correlation improved (r = 0.82). Absence of AB's or low AB counts (less than 1 AB/ml) in BAL corresponded in about 70% of cases to concentrations of less than 1,000 AB/gm and in 100% of cases to concentrations less than 10,000 AB/gm. In subjects with BAL containing more than 1 AB/ml, the lung tissues of 85% contained more than 1,000 AB/gm and the tissues of 44% contained more than 10,000 AB/gm. Above 10 AB/ml BAL, all lung tissues contained more than 10,000 AB/gm. Since lung tissue is not readily available in patients undergoing assessment of their asbestos exposure, BAL fluid analysis seems to be a useful tool to evaluate lung AB concentrations. This technique cannot be performed, however, in patients with severe lung impairment which does not allow sufficient recovery of BAL fluid.