We studied leukocyte chemotaxis triggered by a local injection of mycobacteria (Mycobacterium avium and M. smegmatis) in BALB/c and C57BL/6 mice. Our experimental model consisted of the induction of a subcutaneous air pouch in the dorsal area of mice and inoculation 6 days later of 10(8) CFU of myocobacteria. Inflammatory exudates were harvested from the air pouch cavities 15, 30, and 45 min after the injection of the inocula. Injection of the microorganisms resulted in the migration of an elevated number of eosinophilic granulocytes into the inflammatory cavities. At 30 min after the inoculation of the mycobacteria, the air pouches contained between (3.9 +/- 0.3) x 10(5) (M. avium) and (3.3 +/- 0.3) x 10(5) (M. smegmatis) eosinophils, corresponding to more than one-third (41.4 to 38.3%) of the leukocytes present in the inflammatory cavities. Less than one-half of the eosinophils were attracted to the air pouches when the same number of heat-killed mycobacteria were inoculated [(1.3 +/- 0.2) x 10(5) cells for M. avium and (1.5 +/- 0.2) x 10(5) cells for M. smegmatis]. Injection of gram-negative bacteria (Escherichia coli), of latex beads, or of casein resulted in the attraction of inflammatory eosinophils in numbers that were comparable to those attracted by the heat-killed mycobacteria. Our data document the fact that live mycobacteria exert a rapid chemotactic effect on eosinophils. We therefore postulate that mycobacteria either contain or induce the production of an eosinophilotactic factor. Because this chemotactic effect occurs during the acute inflammatory response to mycobacteria, it cannot be due to the formation of immune complexes (a major infection-associated chemotactic factor for eosinophils). The attracted eosinophils had an important role in the local phagocytosis of mycobacteria, as indicated by our finding, derived from thin-section electron microscopy quantifications, that at 30 min after M. avium inoculation the inflammatory exudates contained (2.2 +/- 0.5) x 10(5) mycobacterium-bearing eosinophils (corresponding to 57% of the total eosinophils), as compared with (2.1 +/- 0.1) x 10(5) neutrophils and (1.5 +/- 0.2) x 10(5) macrophages with ingested bacilli. We conclude that mycobacteria induce the attraction of eosinophils to inflammatory sites and that these granulocytes have the capacity to phagocytize these bacilli in situ.