The ability of inhaled asbestos to produce asbestosis, lung cancer, and mesothelioma in both humans and animals is well established, and asbestos exposures in the occupational and general community environment are recognized as significant hazards. However, it has not been possible to establish realistic and credible dose-response relationships, primarily because of our inability to define which constituents of the aerosols produce or initiate the pathological responses. It is generally acknowledged that the responses are associated with the fibers rather than the nonfibrous silicate mineral of the same chemical composition. Available data from experimental studies in animals exposed by injection and inhalation to fibers of defined size distributions are reviewed, alone with data from studies of fiber distributions in lungs of exposed humans in relation to the effects associated with the retained fibers. It is concluded that asbestosis is most closely related to the surface area of retained fibers, that mesothelioma is most closely associated with numbers of fibers longer than approximately 5 microns and thinner than approximately 0.1 micron, and that lung cancer is most closely associated with fibers longer than approximately 10 microns and thicker than approximately 0.15 micron. The implications of these conclusions on methods for fiber sampling and analyses are discussed.