Asbestos bodies (AB) were counted by light microscopy in bronchoalveolar lavage (BAL) fluid obtained from 563 subjects. The presence of AB was found to reflect occupational exposure to asbestos and was rarely found in unexposed control subjects at concentrations above 1/ml of fluid (6.9% of white collar workers and 17.8% of blue collar workers). The overlap of results observed between subjects with definite exposure and those without underlines the difficulty in assessing exposure by questioning alone, which leads to underestimations or even overestimations of the risk. The highest counts (log mean, 120.5 AB/ml; range, 0 to 42,600) were found in patients with radiologic evidence of asbestosis, most likely reflecting the known association of this disease with retention of large amounts of long amphiboles, rather than in patients with pleural disease. A considerable overlap of results was also observed between groups with different diseases or without any apparent disease. Apart from uncertainties in the radiologic diagnosis, this may be explained by differences in latency since first exposure, in individual response to asbestos inhalation, or in pathogenic properties of different asbestos types. Because the presence of AB in BAL fluid appears to be a marker of exposure and not of disease, AB are more likely to be detected in patients presenting with asbestos-related diseases but in whom exposure is not confirmed by the occupational history (65 of 78 cases).